JP4740828B2 - Information processing apparatus and information processing system - Google Patents

Description

  The present invention relates to an information processing apparatus that is connected to a network and processes packets transferred through the network. More specifically, a packet is received, a different process is executed according to the communication state between the hosts that are transmitting and receiving the packet, data stored inside the apparatus is changed, a new packet is generated, and the external The present invention relates to an information processing apparatus and an information processing system that provide a server service such as a database.

  With the parallel development of broadcasting / communication fusion and ubiquitous computing, the computing performance required from both the backend and access is rapidly expanding. Therefore, appliances that distribute server services (database (Data Base: DB) transactions, anti-abnormal communication protection, etc.) using ubiquitous devices (for example, Radio Frequency Identification (RFID), sensors, cameras, etc.) to the edge network are distributed. It is necessary to realize and improve the computing performance and response speed of the server service.

  The appliance mentioned here realizes anti-abnormal communication protection and database functions.

  Anti-abnormal communication defense detects and eliminates abnormal packets from attackers.

  The database analyzes a large amount of data update packets received from a large number of distributed network terminals such as RFID / stationary sensors / video sensors, aggregates the data in the cache, uploads the data, and stores the cache data in the server Automatically upload to. Also, SQL analysis of a data request packet from a client such as a mobile terminal, XML translation of the data from the cache, download of the data, and automatic download of the server data cache.

  The usage method of the sensor / RFID differs for each user. The contents of the communication packet differ depending on the type of sensor / RFID, the application used, the installation location, and the time such as day and night. Furthermore, the communication state between the terminals that transmit and receive packets varies depending on the TCP congestion / transition state, the L7 protocol in use, the command being executed, and the command execution state.

  Therefore, the functions required for the appliance differ for each user / application / location / time / execution state. In addition, appliances equipped with the above functions need to be distributed and arranged in an edge network as a small server or a board with a built-in communication device in order to improve the efficiency of arithmetic processing. For this reason, a processor for the appliance is required to have a capability of performing various processing for each packet based on the communication state between the terminals in addition to the performance of high-speed packet calculation with small power saving.

  Conventionally, as a device connected to a network and realizing a server service such as a database, a device in which a plurality of general-purpose processors are connected in a cluster (for example, see Non-Patent Documents 1 and 2) is known. In this apparatus, a method of increasing processing capability by distributing processing requests among a plurality of processors is used.

An apparatus that realizes protocol processing on a network using a reconfigurable processor such as an FPGA is known. This apparatus can change the processing content if the frequency is low (see, for example, Non-Patent Document 3).
Haran Boral, David J. DeWitt, “Processor allocation strategies for multiprocessor database machines”, ACM Transactions on Database Systems, June 1981, Volume 6, Issue 2, p. 227-254 Domenico Sacca, Gio Wiederhold, “Database partitioning in a cluster of processors”, ACM Transactions on Database Systems, Volume 10, Issue 1, p. 29-56, March 1985 Ilija Hadzic, JM Smith, “Balancing performance and flexibility with hardware support for network architectures”, ACM Transaction on Computer Systems, November 2003, Volume 21, Issue 4, p. 375-411

  First, an apparatus for realizing a server service such as a database using a conventional general-purpose processor will be described with reference to FIGS.

  In a conventional apparatus using a general-purpose processor, the function of the arithmetic unit 1802 in the processor is switched for each type of program instruction 1801 loaded in the cache memory in the processor (see FIG. 18).

  The arithmetic unit is very multifunctional and can handle complicated and diverse instructions, but the circuit utilization rate is low, such as many of the circuit-mounted transistors are unused during instruction execution. In particular, in database transactions, comparison processing with constraints on the context of data loaded from memory is repeated with large-volume data input / output. The rate is likely to drop and it is difficult to improve performance.

  Further, communication data between a conventional general-purpose processor and a communication line is input / output via a memory 1905 as shown in FIG. Packets received by the network I / F (interface) 1906 from the communication line 1907 are accumulated in the network I / F buffer area 1909 in the memory 1905 via the PCI bus 1904, the PCI bridge 1903, and the external controller 1902. .

  The CPU 1901 reads the accumulated data one by one, executes necessary processing, and accumulates newly generated packets in the network I / F buffer area 1909 again. The network I / F 1906 reads a newly generated packet from the network I / F buffer area 1909 and transmits it to the communication line 1907.

  Therefore, a memory bandwidth corresponding to four times the line bandwidth is consumed (1908).

  In addition to the input / output of communication data, various calculation data are input / output between the CPU and the memory. For this reason, in this method, there is a problem that the CPU idle time for waiting for a free time in the memory band increases and the calculation performance is lowered.

  Due to the above problems, conventionally, a technique is used in which a plurality of general-purpose processors are connected in a cluster, and processing requests are distributed among the plurality of processors to increase the processing capability. On the other hand, the general-purpose processor does not consume high power in order to obtain calculation performance at a high frequency, and it is difficult to save power in the appliance. Moreover, since a fan for waste heat generation of a huge amount of heat is required, there is a problem that it is difficult to downsize the apparatus.

  Furthermore, a conventional reconfigurable processor such as an FPGA realizes high calculation performance with low power consumption, but it takes a long time to change the mounting logic, and the mounting logic cannot be changed frequently. Therefore, conventionally, it has been used for an apparatus that executes a relatively simple process in a low layer such as a protocol process in a communication layer with little change in processing content.

  On the other hand, when using a conventional reconfigurable processor for an appliance that provides server services for application layers such as databases, the communication state between hosts that are sending and receiving packets (transition layer transition / congestion state) In order to execute different processing for each packet depending on the type of application layer protocol used, the type of command being executed, the execution status of the command (how far the file has been sent / received, etc.) Need to change frequently. However, in the conventional reconfigurable processor, the mounting logic cannot be changed frequently, and eventually all the logic necessary for the processing is mounted in advance.

  Therefore, there is a problem that the circuit scale and the number of necessary chips increase, and the apparatus scale and power consumption increase.

  On the other hand, the Dynamic Reconfigurable Processor (DRP) is equipped with an arithmetic matrix in which a large number of small execution units (Execution Elements: EXE) are connected by selectors, and realizes high arithmetic performance equivalent to ASIC by pipeline processing. At the same time, by changing the execution method and connection method of EXE with a minimum of one clock, a high flexibility that can change the mounting logic in a short time is realized.

  DRP is a calculation matrix core operation that does not increase the operating frequency, such as executing simple instructions at once with multiple arithmetic units, or executing complex instructions by combining multiple arithmetic units. By increasing the degree of parallelism, high computing performance is achieved with low power consumption. For this reason, DRP is considered promising as a processor for appliances that simultaneously requires the performance of high-speed packet computation with small power savings and the flexibility to update algorithms. On the other hand, it has a feature that the obtained processor performance varies greatly depending on the software design, circuit reconfiguration method, and device configuration.

  When this dynamic reconfiguration processor is used in a device for application layer server services such as a database, the communication state (transition layer congestion / congestion state of the transport layer protocol) It is necessary to change the installed logic according to the type of application layer protocol being used, the type of command being executed, the progress of command execution (how far the file has been sent and received, etc.), and execute different processing for each packet. There is.

  The change of the mounting logic is started by a reconfiguration trigger generated by the arithmetic unit group inside the arithmetic matrix. Therefore, the information necessary for generating the reconfiguration trigger (communication status between each host transmitting and receiving received packets) is generated in advance outside the calculation matrix and directly input to the calculation matrix. It is thought that the logic can be changed quickly.

  The present invention is equipped with a dynamic reconfigurable processor that uses direct input / output of communication data not via a memory in order to solve the problems (upsizing and power consumption increase) of conventional general-purpose processors and devices using FPGA. In the device, both the high-speed calculation performance and the high-speed change of the mounted logic can be achieved by changing the mounting logic of the operating matrix in the dynamic reconfigurable processor based on the communication state between the hosts generated beforehand outside the operating matrix. An object of the present invention is to provide a small and power-saving information processing apparatus that realizes application layer server services such as databases with high performance.

  According to an embodiment of the present invention, there is provided an information processing apparatus that is connected to a network, receives a packet transferred through the network, and executes processing according to a communication state between hosts that transmit and receive the packet, An interface for transmitting and receiving packets, a transition / congestion state of a transport layer protocol between the hosts, a type of an application layer protocol, a type of command being executed by the application layer protocol and a type of an argument of the command, A communication state storage unit that stores a communication state including a progress state of execution as a communication state table, and is connected to the interface and the communication state storage unit, reads the communication state from the communication state storage unit, and Change the communication status issued, and A communication state changing unit that writes a state in the communication state storage unit and outputs the changed communication state; and a computing unit that is connected to the communication state changing unit and has a plurality of functions. The wiring logic is changed to change the mounting logic, and an arithmetic matrix that processes the received packet based on the communication state output by the communication state changing unit, and is connected to the arithmetic matrix, A circuit configuration accumulation buffer for storing configuration information for specifying a function and mutual wiring, and the communication state changing unit transmits and receives the packet based on information included in the packet received by the interface The communication state between the hosts is read from the communication state storage unit, and the information included in the received packet and the read communication And changing the read communication state based on the contents of the state, writing the changed communication state to the communication state storage unit, and outputting the changed communication state to the arithmetic matrix, The calculation matrix acquires one piece of configuration information from the circuit configuration accumulation buffer based on the communication state output from the communication state change unit, and based on the acquired configuration information, the function of the arithmetic unit group The wiring is reconfigured, the mounting logic inside the operation matrix is changed, and the process corresponding to the communication state is executed on the received packet.

  According to the present invention, it is possible to provide an information processing apparatus that realizes a high-speed reconfiguration of an operation matrix-mounted logic, which is necessary for realizing a server service of an application layer such as a database with a small size, power saving, and high performance.

  Embodiments of the present invention will be described below with reference to the drawings.

  FIG. 1 is a configuration block diagram of an apparatus 200 according to an embodiment of the present invention.

  The apparatus (information processing apparatus) 200 includes a dynamic reconfiguration processor 201, an external controller 202, a memory 205, a network I / F-i (i = 1 to N) 203 (203-1 to 203-N), and The communication line connection unit 204 (204-1 to 204-N) and the switch 211 are configured.

  The device 200 is connected to the network. The network I / F 203 receives the packet 210 (210-1 to 210-N) from the network. The switch 211 transfers the received packet to the external controller 202 or another network I / F 203 based on the destination of the packet. Further, the switch 211 transmits the packet 209 (209-1 to 209-N) transferred from the external controller 202 to the network I / F 203 to the network via the network I / F 203.

  The external controller 202 changes the communication state of the device 200 based on the content of the packet transmitted from the switch 211.

  The dynamic reconfiguration processor 201 includes a processor 206, an operation matrix 207, and an internal controller 208. The dynamic reconfiguration processor 201 is connected to the memory 205. The dynamic reconfiguration processor 201 changes the configuration of the calculation matrix 207 based on the communication state of the external controller 202, and the changed calculation matrix 207 processes the packet transmitted from the external controller 202.

  The memory 205 stores data used by the dynamic reconfiguration processor 201 to change the configuration of the operation matrix 207. The memory 205 can also store data processed by the external controller 202 and the dynamic reconfiguration processor 201.

  As will be described later with reference to FIG. 5, when the apparatus 200 is arranged in the front stage of the server or built in the server, the server 504 is directly connected to the switch 211. A packet destined for the server 504 is directly transmitted to the server 504 by the switch 211.

  FIG. 2 is a detailed configuration block diagram of the dynamic reconfiguration processor 201 included in the apparatus 200.

  The dynamic reconfiguration processor 201 includes a circuit configuration accumulation buffer 101, an autonomous reconfiguration mechanism 109, a processor 206, a calculation matrix 207, and an internal controller 208.

  The processor 206 is a conventional general-purpose processor that executes processing to be executed by the apparatus 200 and also executes processing related to reconfiguration of the operation matrix 207.

  The arithmetic matrix 207 is equipped with a plurality of small arithmetic units whose functions and wirings are variable, and the circuit configuration inside the arithmetic matrix is changed by reading the circuit configuration stored in the circuit configuration storage buffer 101 in advance. it can. The autonomous reconfiguration mechanism 109 acquires the circuit configuration stored in the circuit configuration storage buffer 207 based on the request from the calculation matrix 207 and sends it to the calculation matrix 207.

  The internal controller 208 includes a bus switch 301 that connects the processor 206 or the operation matrix 207 to each interface. The bus switch 301 includes a PCI I / F 302 for PCI bus connection, a DRAM I / F 303 for external memory access, a DMA controller 304 for DMA transfer, and another I / F 305 for connecting to other interfaces. Consists of including.

  FIG. 3 is an explanatory diagram showing a change in the mounting logic of the operation matrix 207 of the dynamic reconfiguration processor 201.

  The arithmetic matrix 207 is equipped with a large number of small arithmetic units whose functions and wirings are variable, and by reading the circuit configuration stored in the circuit configuration storage buffer 101, the function and wiring for each small arithmetic unit are read. And the mounting logic is changed.

  In the example of FIG. 3, the processing contents (401 to 403) change with time, and the mounting logic (404 to 406) of the calculation matrix 207 is changed with the change of the processing contents.

For example, when the processing content is TCP control 401, the arithmetic matrix 207 changes the function of the small arithmetic unit to an adder, a multiplier, etc., reconfigures the wiring of these arithmetic units, and performs the TCP control processing. It changes to the mounting logic 404 suitable for. When the processing content is changed from the TCP control 401 to the data reply 402, the processing logic is similarly changed to the mounting logic 405 suitable for the data reply process. In the same way, change to the mounting logic suitable for the processing contents.
In this way, the operation matrix 207 realizes high-speed operation performance by parallel operation and high flexibility that can change the mounted logic in a short time by changing to the mounted logic suitable for the processing content.

  FIG. 4 is an explanatory diagram of an example of an information processing system including the device 200.

  Device 200 is located in edge network 502 between access network 501 and backend system 503.

  In the back-end system 503, a server 504 and a storage system 505 are arranged. The storage system 505 stores data to be processed by the server 504 (522).

  The access network 501 includes an RFID, a stationary sensor, a video sensor, a mobile terminal, and the like. In addition, an attacker who transmits unauthorized access that the server 504 does not expect may be included.

  Upon receiving the data update packet (508, 509, 510) from the RFID, stationary sensor, or video sensor of the access network 501, the apparatus 200 stores the data update packet (508, 509, 510) in the apparatus 200 based on the SQL command included in the received packet. Update data. Further, based on the format such as HTTP / XML included in the received packet, an update result notification and an action instruction packet (511, 512, 513) are returned.

  In addition, when the data request packet 514 is received from the mobile terminal of the access network 501, the data stored in the apparatus is searched based on a command (SQL command or the like) included in the received packet, and the HTTP / The retrieved data is returned as a request data reply packet 515 based on a format such as XML.

  When the abnormal packet 516 by the attacker is received, it is determined that the communication is abnormal, and the abnormal packet is discarded (507) without being transmitted to the server 504 (517).

  The data stored in the apparatus 200 is uploaded to the server (518, 520) or downloaded from the server (519, 521), and the stored data in the apparatus 200 is always kept up-to-date.

  FIG. 5 is an explanatory diagram of another example of the information processing system including the apparatus 200.

  In FIG. 4 described above, the device 200 is arranged in the edge network 502. However, in this example, the device 200 is arranged in front of the server 504 in the back-end system 603. Note that the apparatus 200 may be built in the server 504. When the device 200 is built in the server 504, the server 212 is directly connected to the switch 211 of the device 200.

  Even when configured as shown in FIG. 5, the processing of the apparatus 200 is the same as the processing described in FIG.

  Next, operations of the dynamic reconfiguration processor 201, the external controller 202, and the memory 205 of the apparatus 200 will be described.

  FIG. 6 is a detailed block diagram of the dynamic reconfiguration processor 201, the external controller 202, and the memory 205.

  The external controller 202 includes a processing packet determination unit 106, a communication state control unit 104, a timer 160, a filter unit 124, a timing adjustment unit 105, a communication state classification unit 103, a delay adjustment unit 118, and an aggregation unit 128. It is comprised including.

  The processing packet determination unit 106 determines whether the received packet is a packet to be processed. The communication state control unit 104 accumulates or updates the current communication state with packets transmitted and received between networks. The timer 160 generates the current time. The filter unit 124 discards the packet determined to be an abnormal packet. The timing adjustment unit 105 adjusts the timing for sending the updated communication state and packet to the dynamic reconfiguration processor 207. The communication state classification unit 103 classifies the communication state. The delay adjustment unit 118 adjusts the delay generated in the communication state and the packet based on the communication state classification result and the reconfiguration completion notification from the operation matrix 207. The aggregation unit 128 aggregates and outputs the packets.

  The communication state control unit 104 includes a communication state table 126 stored in a communication state storage unit that stores a communication state, and a communication state change unit 125 that reads or changes the communication state.

  The timing adjustment unit 105 includes a communication state buffer 121 that temporarily stores a changed communication state, a packet buffer 122 that temporarily stores a packet, and whether or not a packet does not require processing for each packet. And a communication state reading determination unit 123 that reads the communication state when it is determined that processing is necessary.

  The communication state classification unit 103 includes a communication state determination unit 119 that classifies communication states and a packet determination unit 120 that classifies packets.

  The dynamic reconfiguration processor 201 includes a processor 206, an operation matrix 207, an internal controller 208, a circuit configuration accumulation buffer 101, and an autonomous reconfiguration mechanism 109.

  The circuit configuration storage buffer 101 stores a plurality of pieces of configuration information 108 (108-1 to 108-3) for reconfiguring the calculation matrix 207. The autonomous reconfiguration mechanism 109 selects the configuration information 108 (108-1 to 108-3) from the circuit configuration accumulation buffer 101 based on the reconfiguration trigger generated by the arithmetic matrix 207, and the selected configuration information 108 is stored in the selected configuration information 108. Transfer to the operation matrix 207.

  The memory 205 stores a circuit configuration accumulation area 117 and a pointer area 157.

  The circuit configuration storage area 117 records configuration information that cannot be stored in the circuit configuration storage buffer 101. The pointer area 157 records an address pointer to the configuration information in the circuit configuration storage area 117.

  The processor 206 executes the circuit configuration change module 116. The circuit configuration change module 116 reads the configuration information from the circuit configuration storage area 117 based on the reconfiguration trigger generated by the calculation matrix 207 and writes the read configuration information in the circuit configuration storage buffer 101. Then, the configuration information is transferred to the calculation matrix 207.

  The arithmetic matrix 207 includes an autonomous reconfiguration trigger generating arithmetic unit group 111, an external reconfiguration trigger generating arithmetic unit group 112, a server service providing arithmetic unit group 113, and a buffer read determining arithmetic unit group 114. Consists of. However, the configuration of the computing unit group included by the reconfiguration may be changed according to a predetermined condition (for example, time). For example, a time zone configured to include an autonomous reconfiguration trigger generation computing unit 111 and an external reconstruction trigger generation computing unit 112, a server service providing computing unit 113, and a buffer read determination calculation The time zone including the instrument group 114 may be minutes.

  The autonomous reconfiguration trigger generation computing unit 111 outputs a reconfiguration trigger 142 to the autonomous reconfiguration mechanism 109 based on the communication state classification result. The external reconfiguration trigger generation computing unit group 112 outputs the reconfiguration trigger 141 to the circuit configuration change module 116 based on the communication state classification result. Based on the communication state and the packet, the server service providing computing unit group 113 changes the data stored in the memory 205, generates a new packet, and externally transmits the packet. The buffer read determination computing unit group 114 determines whether or not all the packets newly generated by the server service providing computing unit group 113 have been transmitted. The data is output to the timing adjustment unit 105.

  The circuit configuration storage buffer 101 preferentially stores configuration information 108 that is frequently used by the operation matrix 207. On the other hand, the configuration information 108 that is used less frequently is stored in the circuit configuration storage area 117 of the memory 205. The autonomous reconfiguration trigger generating arithmetic unit group 111 and the external reconfiguration trigger generating arithmetic unit group 112 indicate whether the circuit configuration for generating the reconfiguration trigger is stored in the circuit configuration storage buffer 101 or stored in the memory 202. Either the autonomous reconfiguration trigger generation computing unit 111 or the external reconfiguration trigger generation computing unit 112 knows in advance and generates a reconfiguration trigger at the corresponding reconstruction trigger generation timing.

  Hereinafter, the operation of each unit will be described in detail.

  When receiving a packet from the outside (130), the processing packet determination unit 106 refers to the packet header (P.H. 716, see FIG. 7) and determines whether the packet is to be processed. If it is determined that the packet is to be processed, the received packet is output to the communication state changing unit 125 and the filter unit 124 (131). If the packet does not need to be processed, it is output to the aggregation unit 128 (148).

  FIG. 7 is an explanatory diagram showing an example of a format of a packet received by the processing packet determination unit 106 from the outside.

  The packet 130 includes an InLine 701, an OutLine 702, an SMAC 703, a DMAC 704, a Proto 705, a SIP 706, a DIP 707, a Sport 708, a Dport 709, a TCP Flag 710, a PSEQ 711, a PACK 712, a Various Header 713, an Other Header 714, It is comprised including. In this embodiment, SIP 706, DIP 707, Sport 708, and Dport 709 are designated as P.P. H. (Packet Header) 716.

  InLine 701 stores an input line number that is an identification number of a line (network I / F) to which a packet is input. OutLine 702 stores an output line number that is an identification number of a line (network I / F) that outputs a packet. The SMAC 703 stores a source MAC address that is a source address of the data link layer. The DMAC 704 stores a destination MAC address that is a destination address. The Proto 705 stores a network layer protocol. The SIP 706 stores a transmission source address, that is, a transmission source IP address that is the address of the host on the transmission side. The DIP 707 stores a destination address, that is, a destination IP address that is an address of a host on the receiving side. The SPORT 708 stores a TCP transmission source port. The DPORT 709 stores a TCP destination port. The TCP Flag 710 stores a TCP flag. The PSEQ 711 stores a transmission sequence number (SEQ number). The PACK 712 stores a reception sequence number (ACK number). The OtherHeader 713 stores other IP / TCP header data. Various commands 714 store application layer (application layer) commands. The Payload 715 stores data other than the packet header and various commands.

  When the packet is input from the processing packet determining unit 106, the communication state changing unit 125 receives the P.P. H. Based on 716, the communication status table 126 is referenced to determine whether a corresponding communication status exists. If there is a corresponding communication state, it is read and changed. If there is no corresponding communication state, a new entry is created in the communication state table.

  FIG. 8A is an explanatory diagram of an example of the communication state table 126.

  The communication state table 126 includes n entries 801 (801-1 to 801-n).

  The entry 801 includes SIP 802, DIP 803, SPORT 804, DPORT 805, TSEQ 806, TACK 807, WIN 808, SWIN 809, TIME 810, STATE 811, ID 812, and ADD 813.

  The SIP 802 records a transmission source address, that is, a transmission source IP address that is an address of a transmission-side host. The DIP 803 records a destination address, that is, a destination IP address that is the address of the host on the receiving side. The SPORT 804 records the TCP transmission source port. The DPORT 805 records the TCP destination port. TSEQ 806 records the sequence number of the receiving host. TACK 807 records the sequence number of the sending host. WIN 808 records the congestion window size of the TCP connection. The SWIN 809 records a transmitted window size representing the transmitted data size. The TIME 810 records the latest time when the packet related to the entry is received. The STATE 811 stores the state of the TCP connection, the presence / absence and type of an application layer protocol, a command, an argument, and the like. ID812 records the ID number of the transmitted packet. The ADD 813 records the address pointer of the memory 205 in which the end of the ACK received data is stored. In this embodiment, SIP 802, DIP 803, SPORT 804, and DPORT 805 are collectively referred to as T.264. H. (Table Header) 814.

  FIG. 8B is an explanatory diagram illustrating an example of contents stored in the STATE 811.

  The STATE 811 stores information indicating the state of the TCP layer and application layer information.

  Specifically, as information indicating the state of the TCP layer, information (OPEN / CLOSE) 820 indicating the start or stop of the TCP connection, information (FULL / HALF) 821 indicating the establishment or non-establishment of the TCP connection, and TCP Information indicating the congestion state of the connection (Slow Start / Congestion Avidance (hereinafter referred to as Cong. Avoid.) / Fast Recovery) 822 is stored.

  Also, as application layer information, the presence / absence and type of application layer protocol (nothing / HTTP / TELNET / FTP, etc.) 823, the command being executed by the application layer protocol and its arguments and types (GET / POST, SELECT / INSERT / DELETE, etc.) 824 and the file transmission / reception status (start, start, active, passive, etc.) 825 being executed by the command are recorded.

  FIG. 9 is a flowchart of processing of the communication state changing unit 125.

  This flowchart is started when the communication state changing unit 125 receives a packet (step 901).

  When the packet 131 is input from the processing packet determining unit 106, the communication state changing unit 125 determines whether the TCP Flag 710 of the received packet is SYN (step 902).

  If the TCP Flag 710 is SYN in step 902 (step 903), the communication state changing unit 125 reads the P.P. H. T. 716 in accord with 716. H. An entry 801 including 814 is searched (step 904).

  If there is no matching entry (step 905), the current time 161 output by the timer 160 is acquired, and the difference between the latest time 810 at which the packet is received and the current time 161 is compared. As a result of the comparison, an entry 801 whose difference is equal to or larger than a predetermined threshold Thr is searched (step 974). When an entry 801 having a difference equal to or larger than a predetermined threshold Thr is searched (step 907), the searched entry 801 is initialized. The T. of the entry 801 H. 814 of the packet that received 814. H. Change to 716. Further, STATE 811 of this entry 801 is changed to the “HALF” and “OPEN” states. Further, the TSEQ 806 of the entry 801 is changed to a random value. Further, the TACK 807 of the entry 801 is changed to a value obtained by adding 1 to the PSEQ 711 (step 975).

  That is, when a SYN packet that is a session start request is received, the entry 801 of the communication state table 126 is initialized and newly created.

  In step 904, the P.P. H. T. 716 consistent with H. If there is an entry including 814 (step 906), the communication state changing unit 125 obtains the current time 161 output by the timer 160 and compares the difference between the latest time 810 when the packet is received and the current time 161. Then, it is determined whether or not the difference is greater than or equal to a predetermined threshold value Thr (step 908).

  If the difference is greater than or equal to the threshold value Thr (step 909), the communication state changing unit 125 executes the process of step 975. If the difference is not greater than or equal to the threshold Thr (step 910), the communication state of the read entry 801 is not changed, and the filter unit 124 is instructed to discard the received packet (step 976).

  That is, the same P.I. H. When the SYN packet having 714 is retransmitted at an interval shorter than the threshold value Thr, the communication state changing unit 125 determines that the packet is an abnormal packet and discards the packet.

  In step 902, when the TCP Flag 710 is not SYN (step 911), the communication state changing unit 125 reads the P.P of the received packet from the communication state table 126. H. T. 716 consistent with H. An entry 801 including 814 is searched (step 912). P. of received packet. H. T. 716 in accord with 716. H. If there is no entry 801 including 814 (step 913), the communication state changing unit 125 instructs the filter unit 124 to discard the received packet without reading the communication state (step 976).

  That is, when the TCP Flag 710 receives a packet other than SYN in a state where the SYN packet has never been received (a state where the start of the session is not requested), the communication state changing unit 125 regards the packet as an abnormal packet. Judge and discard the packet.

  In step 912, the P.P. H. T. 716 in accord with 716. H. If there is an entry including 814 (step 914), it is determined whether the TCP Flag 710 of the packet is FIN-ACK (step 915).

  If the TCP Flag 710 is FIN-ACK in Step 915 (Step 916), the communication state changing unit 125 receives the entry 801 in which the difference between the PACK 712 value and the TSEQ 806 value of the received packet is 0, that is, reception. It is determined whether or not there is an entry 801 including TSEQ 806 that matches the value of PACK 712 of the received packet (step 917).

  If there is a matching entry 801 (step 918), the communication state changing unit 125 changes the STATE 811 of the entry 801 to the CLOSE state. Also, the TSEQ 806 of this entry 801 is changed to PACK 712. Further, the TACK 807 of the entry 801 is changed to a value obtained by adding 1 to the PSEQ 711 (step 977).

  If there is no matching entry 801 in step 918 (step 919), the filter unit 124 is instructed to discard the packet without reading the communication state of the entry 801 (step 978).

  That is, when the value of PACK 712 included in the received SYN-ACK packet after receiving the SYN packet once is different from the value of TSEQ 806 of the entry already stored in the communication state table 126, the communication state changing unit 125 The packet is determined as an abnormal packet, and the packet is discarded.

  If the TCP Flag 710 is not FIN-ACK in Step 915 (Step 920), it is determined whether or not the TCP Flag 710 of the received packet is RST-ACK (Step 921).

  When the TCP Flag 710 is RST-ACK (step 922), the entry 801 in which the difference between the PACK712 value of the received packet and the TSEQ806 value is 0, that is, the TSEQ806 that matches the PACK712 value of the received packet is set. It is determined whether there is an entry 801 to be included (step 923). If there is a matching entry 801 (step 924), the communication state changing unit 125 changes the STATE 811 of this entry 801 to the CLOSE state (step 979). If there is no matching entry 801 (step 925), the filter unit 124 is instructed to discard the packet without changing the communication state described in the read entry 801 (step 980).

  That is, when the value of PACK 712 included in the FIN-ACK packet received after receiving the SYN packet once is different from the value of TSEQ 806 of the entry already stored in the communication state table 126, the communication state changing unit 125 The packet is determined as an abnormal packet, and the packet is discarded.

  If the TCP Flag 710 is not RST-ACK in Step 921 (Step 926), the communication state changing unit 125 determines whether or not the TCP Flag 710 of the received packet is ACK and the STATE 811 of the matching entry 801 is in the HALF OPEN state. It is determined whether or not (step 927).

  If the TCP Flag 710 is ACK and the STATE 811 is in the HALF OPEN state (step 928), the communication state changing unit 125 determines whether the value obtained by subtracting the TSEQ 806 of the entry 801 from the PACK 712 of the received packet is 1. That is, it is determined whether or not the difference between PACK 712 and TSEQ 806 is 1 (step 929). When the difference between PACK 712 and TSEQ 806 is 1 (step 930), the communication state changing unit 125 changes STATE 811 of the entry 801 to the FULL OPEN state. Also, the TSEQ 806 of this entry 801 is changed to PACK 712 (step 981).

  If the difference between PACK 712 and TSEQ 806 is not 1 in step 929 (step 931), the communication state changing unit 125 discards the packet to the filter unit 124 without changing the communication state of the read entry 801. (Step 982).

  That is, when the value of PACK 712 included in the received ACK packet after receiving the SYN packet once is different from the value of TSEQ 806 of the entry already stored in the communication state table 126, the communication state changing unit 125 The packet is determined as an abnormal packet, and the packet is discarded.

  If the TCP Flag 710 is not ACK or the STATE 811 is not in the HALF OPEN state in Step 927 (Step 932), the communication state changing unit 125 determines whether the TCP Flag 710 of the received packet is SYN-ACK, and It is determined whether STATE 811 of entry 801 is in the HALF OPEN state (step 933).

  In step 933, when the TCP Flag 710 is SYN-ACK and the STATE 811 is in the HALF OPEN state (step 934), the communication state changing unit 125 determines that the difference between the PACK 712 of the received packet and the TSEQ 806 of the entry 801 is the difference. It is determined whether or not 1 (step 935).

  If the difference between the PACK 712 and the TSEQ 806 is 1 (step 936), the communication state changing unit 125 changes the STATE 811 of the entry 801 to the FULL OPEN state, and changes the TSEQ 806 of the entry 801 to the PACK 712 (step 983). . If the difference between PACK 712 and TSEQ 806 is not 1 (step 937), the communication state changing unit 125 instructs the filter unit 124 to discard the packet without changing the communication state of the read entry 801 (step 937). 984).

  If the TCP Flag 710 is not SYN-ACK or the STATE 811 is not in the HALF OPEN state in Step 933 (Step 938), the communication state changing unit 125 determines whether the TCP Flag 710 of the packet is PSH-ACK, and It is determined whether STATE 811 of entry 801 is in the FULL OPEN state (step 939).

  When the TCP Flag 710 is PSH-ACK and the STATE 811 is in the FULL OPEN state (step 940), the communication state changing unit 125 matches the value of the PACK 712 of the received packet with the value of the TSEQ 806 in the entry 801. Whether or not (step 941). If the PACK 712 and the TSEQ 806 do not match (step 942), the communication state changing unit 125 instructs the filter unit 124 to discard the packet without changing the communication state of the read entry 801 (step 985). .

  In step 941, when PACK712 and TSEQ806 match (step 943), the communication state changing unit 125 refers to the contents of various commands 714 in the packet. Then, it is determined whether or not the contents of the various commands 714 indicate that the application layer protocol is HTTP and this is a GET command based on HTTP (step 944).

  If the GET command is based on HTTP (step 945), the communication state changing unit 125 changes STATE 811 of the entry 801 to HTTP GET Start, that is, a file transmission / reception start state. Also, the value of TACK 807 in this entry 801 is changed to the sum of the value of PSEQ 711, the byte size of various commands 714, and the byte size of Payload 715. Further, the value of TSEQ 806 of this entry 801 is changed to the value of PACK 712 (step 986).

  In step 944, if the contents of the various commands 714 are not HTTP GET commands (step 946), it is determined whether or not the contents of the various commands 714 are the HTTP application layer protocol and the HTTP POST command. (Step 947). If the command is an HTTP POST command (step 948), the communication state changing unit 125 determines whether the command argument included in the various commands 714 of the received packet is SELECT (step 949).

  If the command argument is SELECT (step 950), the communication state changing unit 125 changes STATE 811 of the entry 801 to HTTP POST SELECT Start, that is, a file transmission / reception start state. Further, the TACK 807 of the entry 801 is changed to the sum of the value of the PSE 711, the byte size of various commands 714, and the byte size of the Payload 715. Further, the value of TSEQ 806 of this entry 801 is changed to the value of PACK 712 (step 987).

  When the command argument is not SELECT (step 951), the communication state changing unit 125 determines whether or not the command argument included in the various commands 714 of the received packet is DELETE (step 952).

  When the command argument is DELETE (step 953), the communication state changing unit 125 changes STATE 811 of the entry 801 to HTTP POST DELETE Start, that is, a file transmission / reception start state. Also, the value of TACK 807 of entry 801 is changed to the sum of the value of PSEQ 711, the byte size of various commands 714, and the byte size of Payload 715. Also, the value of TSEQ 806 of entry 801 is changed to the value of PACK 712 (step 988).

  If the command argument is not DELETE (step 954), the communication state changing unit 125 determines whether or not the command argument included in the various commands 714 of the packet is INSERT (step 955).

  If the command argument is INSERT (step 956), the communication state changing unit 125 changes STATE 811 of the entry 801 to HTTP POST INSERT Start, that is, a file transmission / reception start state. Also, the value of TACK 807 of entry 801 is changed to the sum of the value of PSEQ 711, the byte size of various commands 714, and the byte size of Payload 715. Further, the value of TSEQ 806 of entry 801 is changed to the value of PACK 712 (step 989).

  If the command argument is not INSERT, the process ends. Alternatively, it is determined whether it is another command argument, and processing suitable for the determined argument is executed.

  If it is determined in step 947 that the command is not an HTTP POST command, the process ends. Alternatively, it is determined whether it is another application layer protocol, and processing suitable for the determined protocol is executed.

  In step 939, if the TCP Flag 710 is not PSH-ACK or the STATE 811 is not in the FULL OPEN state (Step 957), the communication state changing unit 125 indicates that the TCP Flag 710 is ACK and the STATE 811 is in the Active state. It is determined whether or not there is (step 958).

  When the TCP Flag 710 is not ACK or the STATE 811 is not in the Active state (Step 960), the communication state changing unit 125 discards the packet to the filter unit 124 without changing the communication state of the read entry 801. (Step 973).

  When the TCP Flag 710 is ACK and the STATE 811 is in the Active state (Step 959), the communication state changing unit 125 determines whether or not the value of the PACK 712 in the entry 801 matches the value of the TSEQ 806 ( Step 961).

  If the value of PACK 712 matches the value of TSEQ 806 (step 963), the communication state changing unit 125 changes the STATE 811 of the entry 801 to the Fast Recovery state. Also, the value of TACK 807 of entry 801 is changed to the sum of the value of PSEQ 711, the byte size of various commands 714, and the byte size of Payload 715. Also, the value of TSEQ 806 of entry 801 is changed to the value of PACK 712 (step 970).

  In step 961, if the value of PACK712 in the received packet does not match the value of TSEQ806 in entry 801 (step 962), the difference between the value of PACK712 in the received packet and the value of TSEQ806 is greater than 0, and congestion occurs. It is determined whether or not the window size is smaller than WIN 808 (step 964).

  If the difference between the value of PACK 712 and the value of TSEQ 806 is 0 or less or the congestion window size WIN 808 or more (step 966), the packet is discarded to the filter unit 124 without changing the communication state of the read entry 801. (Step 973).

  If the difference between the value of PACK 712 and the value of TSEQ 806 is greater than 0 and smaller than the congestion window size WIN 808 (step 965), the congestion window size WIN 808 is set to a predetermined maximum window size, that is, in the network. It is determined whether or not it is larger than ssth, which is a half value of the maximum allowable capacity (step 967).

  When the congestion window size WIN 808 is larger than ssth (step 968), the communication state changing unit 125 changes the STATE 811 of the entry 801 to the Congestion Avidance state. Also, the value of TACK 807 of entry 801 is changed to the sum of the value of PSEQ 711, the byte size of various commands 714, and the byte size of Payload 715. Also, the value of TSEQ 806 of entry 801 is changed to the value of PACK 712 (step 972).

  If the congestion window size WIN 808 is smaller than ssth (step 969), the communication state changing unit 125 changes the STATE 811 of the entry 801 to the Slow Start state. Also, the value of TACK 807 of entry 801 is changed to the sum of the value of PSEQ 711, the byte size of various commands 714, and the byte size of Payload 715. Further, the value of TSEQ 806 of entry 801 is changed to the value of PACK 712 (step 971).

  As described above, the communication state changing unit 125 changes and changes the read communication state 801 based on the information included in the read communication state 801 and the information included in the received packet 130 by the processing illustrated in FIG. The completed communication state 134 is output to the communication state buffer 121 (process 132 in FIG. 6). As a result, the communication state necessary for changing the mounting logic of the operation matrix 207 can be changed according to the change of the communication state accompanying the packet transfer. If there is an abnormality in the combination of the communication state 801 information, the packet 131 information, and the current time 161 generated by the timer 160, the unauthorized access can be eliminated by discarding the packet.

  Note that STATE 811 of the communication state 134 output by the communication state changing unit 125 is a unique value indicating one of the combinations shown in FIG. 8B.

  Returning to FIG. 6, when the filter unit 124 receives the packet discard instruction 133 output in Steps 976, 978, 980, 982, 984, 985, or 973 of FIG. The received packet 131 input from the unit 106 is discarded. The packet 135 that has not been discarded is sent to the packet buffer 122.

  The packet 135 stored in the packet buffer 122 reads and outputs one packet 136 at the timing when the buffer read command 139 output from the buffer read determination arithmetic unit group 114 in the arithmetic matrix 207 is received. If the packet buffer 122 is empty, nothing is done.

  The packet 136 read from the packet buffer 122 is output to the communication state read determination unit 123. The communication state read determining unit 123 determines the P.P. H. Referring to 716, it is determined whether or not the packet is to be processed.

  If the packet is to be processed, the communication state 138 is read from the communication state buffer 121 (step 137), and the packet 107 to be processed is output to the packet determination unit 120 and the delay adjustment unit 118. If it is determined that the packet is not to be processed, the packet 147 that does not need to be processed is output to the aggregation unit 128. Then, the next packet read command 139 is output to the packet buffer 122.

  The communication state determination unit 119 classifies the communication state with reference to the content of the communication state output from the communication state reading unit 123.

  The communication status is classified by referring to a value indicating the start or stop of the TCP connection included in the communication status, and determining the start or stop of the TCP connection, and the TCP connection is started. Classifies the communication state by determining the establishment or non-establishment of the TCP connection with reference to the value indicating the establishment or non-establishment of the TCP connection included in the communication state, and when the TCP connection is established, By referring to the value indicating the congestion state of the TCP connection included in the communication state, classifying the communication state by determining the congestion state of the TCP connection, and determining the presence or absence of the application layer protocol included in the communication state If there is an application layer protocol that classifies the communication state The communication status is classified by determining the type of application layer protocol included in the communication status and the presence of commands and command arguments being executed by the protocol. If there are commands and command arguments, the command and command arguments The communication state is classified by determining whether the progress state of the file transmission / reception being executed is started, incomplete, or completed by the command.

  The classification result of the communication state classified by these values is output to the autonomous reconfiguration trigger generation operation unit group 111 and the external reconfiguration trigger generation operation unit group 112 of the operation matrix 207.

  FIG. 10 shows a group 111 of autonomous reconfiguration trigger generation units in the operation matrix 207 based on the communication state classification result 150 of the communication state determination unit 119 and the packet classification result 151 of the packet determination unit 120, and the external reconfiguration. 6 is a flowchart of processing in which the trigger generation computing unit 112 generates an autonomous reconfiguration trigger 141 or an external reconfiguration trigger 142.

  This processing is executed simultaneously by the autonomous reconfiguration trigger generation computing unit 111 and the external reconstruction trigger generation computing unit 112. If the information necessary for reconfiguration is configuration information stored in the circuit configuration storage buffer 101, the autonomous reconfiguration trigger generation computing unit 111 outputs an autonomous reconfiguration trigger to the autonomous reconfiguration mechanism 109. To do. On the other hand, when the information necessary for reconfiguration is the configuration information stored in the external memory 205, the external reconfiguration trigger generation computing unit 112 sends an external reconfiguration trigger to the circuit configuration change module 116 of the processor 206. Is output.

  When the communication state classification result 150 and the packet classification result 151 are input to the autonomous reconfiguration trigger generation arithmetic unit group 111 and the external reconfiguration trigger generation arithmetic unit group 112 (step 1001), the communication state classification result is obtained. Then, it is determined whether STATE 811 in the communication state is HALF OPEN and the TCP Flag 710 of the packet is SYN (step 1002).

  In step 1002, if the communication state STATE 811 of the communication state is HALF OPEN and the TCP Flag 710 of the packet is SYN (step 1003), the TCP Flag 710 is received for the received packet. A reconfiguration trigger is generated that requests transfer of the configuration information describing the logic for returning the packet changed to SYN-ACK to the operation matrix 207 (step 1040).

  In step 1002, if the communication state STATE 811 of the communication state is HALF OPEN and the TCP Flag 710 of the packet is not SYN (step 1004), the communication state classification result is the communication state of the communication state. It is determined whether STATE 811 in the communication state is CLOSE and TCP Flag 710 of the packet is FIN-ACK (step 1005).

  In step 1005, when the communication state STATE 811 of the communication state is CLOSE and the TCP Flag 710 of the packet is FIN-ACK (step 1006), the TCP flag 710 is received for the received packet. A reconfiguration trigger is generated to request that the configuration information describing the logic for returning the packet changed to FIN-ACK be transferred to the operation matrix 207 (step 1041).

  In step 1005, if the communication state STATE 811 of the communication state is CLOSE and the TCP Flag 710 of the packet is not FIN-ACK in step 1005 (step 1007), then the communication state classification result is Whether the communication state STATE 811 is CLOSE and the packet TCP Flag 710 is RST-ACK is determined (step 1008).

  In step 1008, if the communication state STATE 811 is CLOSE and the TCP Flag 710 of the packet is RST-ACK (step 1009), the TCP flag 710 is set to RST- for the received packet. A reconfiguration trigger is generated that requests that the configuration information describing the logic for returning the packet changed to ACK be transferred to the operation matrix 207 (step 1042).

  In step 1008, when the communication state STATE 811 is CLOSE and the TCP Flag 710 of the packet is not RST-ACK (step 1010), the communication state classification result is that the communication state STATE 811 is the communication state STATE 811. It is determined whether the packet is FULL OPEN and the TCP Flag 710 of the packet is SYN-ACK (step 1011).

  In step 1011, when the communication state STATE 811 is FULL OPEN and the packet TCP Flag 710 is SYN-ACK (step 1012), the communication state classification result indicates that for the received packet, A reconfiguration trigger is generated that requests that the configuration information describing the logic for returning the packet in which the command 714 is changed to the command of the application layer protocol indicated by the STATE 811 in the communication state is transferred to the operation matrix 207 ( Step 1043).

  In step 1011, if the communication state STATE 811 is FULL OPEN and the packet TCP Flag 710 is not SYN-ACK (step 1013), the communication state classification result is the communication state STATE 811. Is GET / POST Active (during file transmission / reception), and it is determined whether the TCP Flag 710 of the packet is ACK (step 1014).

  If the communication state STATE 811 is GET / POST Active (during file transmission / reception) and the packet TCP Flag 710 is ACK in step 1014 (step 1016), the communication state classification is performed. As a result, it is determined whether or not the STATE 811 in the communication state is classified as Slow Start (step 1017).

  In step 1017, when the communication state STATE 811 is Slow Start in step 1017 (step 1018), the configuration information describing the logic for transmitting and receiving packets according to TCP congestion window control at the time of slow start Is generated in the operation matrix 207 (step 1044).

  If the communication state STATE 811 is not the slow start in step 1017 (step 1019), it is determined whether or not the communication state STATE 811 is the Congestion Avidance (step 1019). Step 1020).

  In step 1020, when the communication state STATE 811 is “Consultation Avidance” (step 1021), the configuration information describing the logic for transmitting and receiving packets according to the TCP congestion window control during congestion aviation Is generated in the operation matrix 207 (step 1045).

  In step 1020, when the communication state STATE 811 is not Congestion Avidance (step 1022), the communication state classification result determines whether the communication state STATE 811 is Fast Recovery (step 1022). 1055).

  In step 1055, when the communication state STATE 811 is Fast Recovery (step 1023), the configuration information describing the logic for transmitting and receiving packets according to the TCP congestion window control at the time of Fast Recovery is described. Is generated to request the transfer to the operation matrix 207 (step 1046).

  In step 1014, if the communication state STATE 811 is GET / POST Active, that is, file transmission / reception is performed, and the packet TCP Flag 710 is not ACK (step 1015), the communication state classification result is step 1015. However, it is determined whether or not STATE 811 in the communication state is GET Start, that is, file transmission / reception is started, and the TCP Flag 710 of the packet is classified as PSH-ACK (step 1024).

  In step 1024, when the communication state STATE 811 is HTTP GET Start, that is, file transmission / reception start and the packet TCP Flag 710 is PSH-ACK in step 1024 (step 1025), A reconfiguration trigger is generated to request that the configuration information describing the logic for searching the requested file and returning the search result as a packet is transferred to the operation matrix 207 (step 1047).

  In step 1024, if the communication state STATE 811 is HTTP GET Start, that is, file transmission / reception start and the packet TCP Flag 710 is not PSH-ACK in step 1024 (step 1026), the communication state classification As a result, it is determined whether STATE 811 in the communication state is HTTP POST Start, that is, file transmission / reception start, and the TCP Flag 710 of the packet is PSH-ACK (step 1027).

  In step 1027, when the communication state STATE 811 is HTTP POST Start, that is, file transmission / reception start and the packet TCP Flag 710 is not PSH-ACK (step 1029), the packet is discarded. A reconfiguration trigger is generated that requests to transfer the configuration information describing the logic for the operation matrix 207 (step 1048).

  If the communication state STATE 811 is HTTP POST Start (file transmission / reception start) and the packet TCP Flag 710 is PSH-ACK in step 1027 (step 1028), the communication state classification result is as follows. It is determined whether the result STATE 811 in the communication state is SELECT (step 1052).

  In step 1052, if the communication status STATE 811 is SELECT (step 1031), the logic for searching the data requested by the SELECT command and returning the search result in a packet is described. A reconstruction trigger is generated that requests transfer of the configured information to the operation matrix 207 (step 1049).

  If the communication state STATE 811 is not SELECT in step 1052 (step 1030), the communication state classification result determines whether the communication state STATE 811 is DELETE (step 1053). .

  If the communication state STATE 811 is DELETE in step 1053 (step 1033), the data requested to be deleted by the DELETE command is searched, the searched data is deleted, and the deletion result A reconfiguration trigger is generated that requests transfer of the configuration information describing the logic for returning the packet indicating to the operation matrix 207 (step 1050).

  If the communication state STATE 811 is not DELETE in step 1053 (step 1032), the communication state classification result determines whether the communication state STATE 811 is INSERT (step 1054). .

  If the communication state STATE 811 is INSERT in step 1054 (step 1034), the data requested to be inserted by the INSERT command is written, and a packet indicating the result of insertion by writing is returned. A reconfiguration trigger is generated that requests transfer of the configuration information describing the logic to be performed to the operation matrix 207 (step 1051).

  The autonomous reconfiguration trigger generation operation unit group 111 and the external reconfiguration trigger generation operation unit group 112 in the operation matrix 207 perform the communication state classification result 150 and the packet determination unit of the communication state determination unit 119 by the processing described above. Based on the 120 packet classification results 151, the autonomous reconfiguration trigger 142 or the external reconfiguration trigger 141 is generated. The external reconfiguration trigger generation computing unit 112 generates the external reconfiguration trigger 141 and simultaneously writes the value of the address pointer in which the requested configuration information is written in the pointer area 157 in the memory 205 (process 158).

  In this embodiment, it is determined in step 1014 whether or not the communication state categorization result STATE 811 is GET / POST Active. However, GET and POST are classified separately for each command. Different reconfiguration triggers may be generated. In the present embodiment, an example in which a reconfiguration trigger is generated using both the communication state classification result 150 and the packet classification result 151 has been shown. However, only the communication state classification result 150 is used to generate a reconfiguration trigger. It may be generated.

  When an autonomous reconfiguration trigger 142 is input, the autonomous reconfiguration mechanism 109 receives any of the configuration information 108-1 to 108-3 stored in the circuit configuration accumulation buffer 108 based on the type of the autonomous reconfiguration trigger 142. The selected configuration information 108-k (k = 1 to 3) is transferred to the operation matrix 207 (processing 143). The arithmetic matrix 207 reconfigures the functions and wiring of the internal arithmetic unit group in accordance with the transferred configuration information 108-k, and changes the mounting logic. At this time, the mounting logic of the server service providing computing unit group 113 is mainly changed. The server service providing computing unit group 113 whose mounting logic is changed processes the received packet based on the mounting logic, changes the data accumulated in the apparatus 200, generates a new packet, and is generated Perform external packet transmission.

  When the external reconfiguration trigger 141 is input, the circuit configuration change module 116 executed by the processor 206 reads the value of the address pointer describing the configuration information from the pointer area 157 in the memory 205 (process 159).

  Further, the configuration information is read from the circuit configuration storage area 117 in the memory 205 using the read value of the address pointer (process 149). The read configuration information replaces one of the configuration information 108-k (k = 1 to 3) stored in the circuit configuration accumulation buffer 108 (process 144) and is transferred to the operation matrix 207 (process 143). The arithmetic matrix 207 reconfigures the functions and wiring of the internal arithmetic unit group in accordance with the transferred configuration information 108-k, and changes the mounting logic.

  Note that by reading the configuration information 108 from the circuit configuration accumulation buffer 101, high-speed circuit reconfiguration is possible. On the other hand, since the storage capacity of the circuit configuration storage area 117 is significantly larger than the storage capacity of the circuit configuration storage buffer 108, storing the configuration information in the external memory 205 greatly increases the number of mounted logics that can be changed. To do.

  As described above, according to the processing shown in FIG. 10, the operation matrix 207 changes the mounting logic for each packet based on the classification result 150 of the communication state between hosts generated in advance by the communication state determination unit 119. Further, the autonomous reconfiguration mechanism 109 autonomously reconfigures the mounted logic by the autonomous reconfiguration trigger 142 output by the autonomous reconfiguration mechanism 109. Further, the mounting logic stored in the memory 205 is changed by the external reconfiguration trigger 141 output by the external reconfiguration trigger generation computing unit group 112. Realizes a large increase in the number of mountable logic.

  The delay adjustment unit 118 predicts the timing at which the reconfiguration of the operation matrix 207 ends based on the communication state classification result 150 of the communication state determination unit 119 and the packet classification result 151 of the packet determination unit 120.

  Note that the operation matrix 207 may detect the end of its own reconfiguration process and output a reconfiguration end notification 162 to the delay adjustment unit 118.

  The delay adjustment unit 118 adjusts the delay generated in the packet 107 and the communication state 138 based on the predicted timing or the reconfiguration end notification 162 output from the calculation matrix 207.

  That is, the delay adjustment unit 118 adjusts this delay so that the communication state 138 and the packet 107 are input to the server service providing arithmetic unit group 113 after the change of the mounting logic of the arithmetic matrix 207 is completed. .

  The delay adjustment unit 118 adjusts the delay to be generated based on the communication state classification result 150 of the communication state determination unit 119 and the packet classification result 151 of the packet determination unit 120. The delayed communication state 152 and packet 153 are output to the server service providing computing unit 113. Thereby, the packet 107 and the communication state 138 are input to the calculation matrix 207 at the timing when the reconfiguration of the calculation matrix 207 is completed.

  The server service providing computing unit 113 implements a server service such as a database by changing the data stored in the memory 205 (processing 156) or generating a new packet 146 and transmitting it to the outside. Near the end of processing, a processing end notification 140 is output to the buffer read determination computing unit group 114. In addition, a communication state 155 newly generated when upload / download of data accumulated in the memory 205 is started with the server 504 is output to the communication state change unit 125.

  Further, the server service providing arithmetic unit group 113 includes the transmission / reception sequence number PSEQ711 / PACK712 described in the received packet 153 when the STATE 811 of the input communication state 152 is FULL OPEN, that is, when the TCP connection is established. Based on the transmission / reception sequence number TSEQ 806 / TACK 807 described in the received communication state 152, the congestion window size WIN 808, and the transmitted window size SWIN 809, the STATE 811 is set to Slow Start / Consistency Aviation / Fast Recovery, that is, TCP connection. The communication state 155 in which the congestion state is changed to one of the slow start state / congestion avoidance state / fast recovery state, And it outputs the signal state changing section 125.

  Further, the server service providing arithmetic unit group 113 receives the STATE 811 in the received communication state 152 as FULL OPEN, that is, when the TCP connection is established, and the Slow Start / Congestion Avidance, that is, the congestion state is the slow start state / congestion state. In the avoidance state, based on the transmission size of the newly generated packet, change the values of the congestion window size WIN 808 representing the data size that can be transmitted at one time and the transmitted window size SWIN 809 representing the transmitted data size. Based on the number of newly transmitted packets, the communication state 155 in which the ID number ID 812 of the transmitted packet is changed is output to the communication state changing unit 125.

  Further, the server service providing arithmetic unit group 113, when the received STATE 811 of the communication state 152 is HALF OPEN, that is, when the TCP connection is not established, the communication state 155 in which the ID 812 indicating the ID number of the transmitted packet is initialized Is output to the communication state changing unit 125.

  Further, in the server service providing arithmetic unit group 113, the received STATE 811 in the communication state 152 is Start, that is, the application layer protocol is being executed, and the file transmission / reception being executed by the protocol command is in the start state. In this case, after transmitting the newly generated packet, STATE 811 is changed to Active, that is, the communication state 155 in which the state of file transmission / reception is changed to an incomplete state is output to the communication state change unit 125.

  Further, in the server service providing computing unit 113, STATE 811 in the received communication state 152 is Start / Active, that is, there is a command in which the application layer protocol is being executed, and file transmission / reception being executed by the protocol command is started. In the case of the state or the incomplete state, the communication state 155 in which the address pointer ADD813 of the memory 205 storing the end of the ACK received data is initialized is based on the transmission size of the newly generated packet. Output to the changing unit 125.

  Further, in the server service providing computing unit 113, STATE 811 in the received communication state 152 is Start / Active, that is, there is a command in which the application layer protocol is being executed, and file transmission / reception being executed by the protocol command is started In the state / incomplete state, when there is no packet to be newly generated and transmitted, STATE 811 is Passive, that is, the communication state 155 in which the state of file transmission / reception is changed to complete is output to the communication state change unit 125. .

  Through the processing described above, the communication state can be changed according to the processing content of the server service providing computing unit group 113.

  The communication state changing unit 125 adds the communication state 155 input from the server service providing computing unit group 113 to the communication state table 126.

  The buffer read determination computing unit group 114 outputs whether or not the server service providing computing unit group 113 has transmitted all newly generated packets by the server service providing computing unit group 113 at the end of processing. A determination is made based on the notification 140. When it is determined that all have been transmitted, a buffer read command 139 is output to the packet buffer 122 in the timing adjustment unit 105. By doing so, in the arithmetic matrix 207, after the processing of a certain packet is completed, the processing of another packet is immediately executed. Thereby, the utilization efficiency of the calculation matrix 207 is improved.

  The aggregating unit 128 aggregates the packet 146 from the server service providing arithmetic unit group 113 and the processing unnecessary packet 147 from the communication state read determining unit 123 or the processing unnecessary packet from the processed packet determining unit 106 and aggregates them outside. Packet 154 is transmitted. In this way, by inputting only the packets that need processing to the calculation matrix 207, it is possible to prevent the packets that do not need to be input to the calculation matrix 207.

  Through the processing described above, the dynamic reconfiguration processor and the apparatus using the direct input / output of communication data that does not pass through the external memory are used to perform dynamic reconfiguration based on the communication state between hosts generated in advance outside the processor. It is possible to change the mounting logic of the operation matrix in the processor.

  11 to 17 are sequence diagrams of server services realized by the apparatus 200. FIG.

  FIG. 11 is a sequence diagram in which the apparatus 200 excludes an attacker's abnormal packet.

  The attacker 1100 spoofs the transmission source IP address 706 toward the server 504, and transmits a packet 1101 in which the transmission sequence number PSEQ711 is X, the reception sequence number PACK712 is 0, and the TCP Flag 710 is SYN. The apparatus 200 performs the process of step 975 via step 902, step 904, and step 974 of FIG.

  By the processing in step 975, T.P. H. 814. H. Then, the communication state is changed, that is, STATE 811 is changed to Half Open, TACK 807 is changed to PSEQ 711 + 1, and a random value Y is embedded in the sequence number TSEQ 806 of the receiving host (processing 1102). Furthermore, by the processing of step 1040 in FIG. The operation matrix 207 changed to the logic of returning a packet in which the TCP Flag 710 is set to SYN-ACK returns a SYN-ACK packet 1103 in which the PSEQ 711 is set to the value Y of the TSEQ 806.

  Since the attacker 1100 has disguised the source IP address 706, the attacker 1100 cannot receive the SYN-ACK packet 1103 (processing 1104). Therefore, an appropriate value Z + 1 is set as PACK 712, and a packet 1105 in which TCP Flag 710 is ACK is transmitted. The apparatus 200 determines that the value of PACK 712 is different from the value of TSEQ 806 (Y) +1 in step 929 for the received packet via step 927 in FIG. 9, and discards the packet by the process of step 982 ( Process 1106).

  When the attacker 1100 transmits the SYN packet 1107 again using the same source IP address 706 as that of the SYN packet 1101, the apparatus 200 performs the process of step 910 via step 902 and step 904. When it is determined that the difference between the latest time 133 generated by the timer 160 and the latest time TIME 810 of packet reception in the communication state is smaller than a predetermined threshold Thr, the packet is discarded by the process of step 976 (process 1108). ).

  As described above, the apparatus 200 according to the present embodiment can eliminate abnormal packets due to a SYN flood type DDoS / DoS attack by the process described with reference to FIG.

  FIG. 12 shows a sequence diagram in which the apparatus 200 protects normal packets from the access terminal 1200 such as an RFID / sensor / camera.

  The access terminal 1200 transmits to the server 504 a packet 1201 in which the transmission sequence number PSEQ 711 is X, the reception sequence number PACK 712 is 0, and the TCP Flag 710 is SYN. The apparatus 200 executes the process shown in Step 975 via Step 902, Step 904, and Step 974 in FIG. In the process of step 975, T.P. H. 814. H. Then, the communication state is changed, that is, STATE 811 is changed to HALF OPEN, TACK 807 is changed to PSEQ 711 + 1, and a random value Y is embedded in the sequence number TSEQ 806 of the receiving host (processing 1202). Further, the processing matrix 207 changed to the logic of returning a packet in which the TCP Flag 710 is set to SYN-ACK by the processing of Step 1040 in FIG. 10 returns a SYN-ACK packet 1203 in which the PSEQ 711 is set to the value Y of the TSEQ 806.

  When receiving the SYN-ACK packet 1203, the access terminal 1200 uses the value of the PSEQ 711 (Y) described in the SYN-ACK packet 1203, sets the PACK 712 to Y + 1, and transmits the packet 1204 in which the TCP Flag 710 is ACK. The apparatus 200 determines that the value of ACK 712 matches the value of TSEQ 806 (Y) +1 by the process of step 929 via step 927 (process 1205), and executes the process of step 981. In the process of Step 981, TSEQ806 is changed to PACK712, and the communication state is changed, that is, STATE 811 is changed to FULL OPEN. Thereby, communication can be continued.

  As described above, the apparatus 200 can protect the normal packet from the access terminal 1200 by the processing described with reference to FIG.

  FIG. 13 shows a sequence diagram in which the device 200 that has received the GET command requesting “file A” using the application layer protocol HTTP transmitted from the access terminal 1200 updates the data accumulated in the device. Here, the external memory 205 includes a pointer table 1312 and a file table 1314 in advance.

  When the device 200 receives the packet 1301 describing the GET command requesting the file A using the application layer protocol HTTP, the device 200 executes the processing of step 986 via steps 939, 941 and 944. In step 986, TSEQ806 is changed to PACK712, TACK807 is changed to the sum of PSE711, various commands 714 and Payload 715, and the communication state is changed, that is, STATE 811 is changed to HTTP GET Start. Furthermore, the operation matrix 207 whose mounting logic has been changed by the processing of step 1047 in FIG. 10 searches the pointer table 1312 and determines whether or not the file A is stored in the memory 205 (processing 1313). When the file A is not accumulated, the operation matrix 207 generates a packet 1302 in which the TCP Flag 710 is set to SYN and transmits the packet 1302 to the server 504. Further, the new communication state 155 between the apparatus 200 and the server 504 at this time is output to the communication state changing unit 104.

  When the apparatus 200 receives the SYN-ACK packet 1303 in which the TCP Flag 710 is SYN-ACK from the server 504, the process of step 983 is executed via steps 933 and 935. In the process of step 983, TSEQ806 is changed to PACK712, and the communication state is changed, that is, STATE 811 is changed to FULL OPEN. Further, the operation matrix 207 whose mounting logic has been changed by the processing of Step 1043 in FIG. 10 includes an ACK packet 1304 in which the TCP Flag 710 is ACK, and a packet including a GET command for requesting the file A using the application layer protocol HTTP. 1305 is transmitted. Further, the operation matrix 207 includes a communication state 155 including the changed congestion window size 808, the transmitted window size 809, the ID number ID 812 of the transmitted packet, and the STATE 811 in which the file transmission / reception state is changed to Active. And output to the communication state changing unit 104.

  When the apparatus 200 receives the ACK packet 1306/1308 including the divided data 1315 of the file A from the server 504, the apparatus 200 receives the ACK packet 1306/1308 including the divided data 1315 of the file A, and then goes to steps 970, 971, or 972 via the steps 958, 961, 964, or 967. Execute the process. In the process of step 970, 971 or 972, TSEQ806 is changed to PACK712, TACK807 is changed to the sum of PSEQ711, various commands 714, and Payload715, and the communication state is changed, that is, STATE811 is set to Slow Start / ConsistencyAvailance / FasterRevance Change to one of the following.

  Furthermore, the operation matrix 207 whose mounting logic has been changed by the processing of step 1044, 1045 or 1046 of FIG. 10 accumulates the received divided data 1315 of file A in the file table 1314 (processing 1316). At this time, the TCP Flag 710 returns an ACK packet 1307/1309 which is ACK. Further, the communication state 155 including the changed congestion window size 808, the transmitted window size 809, the ID number ID 812 of the transmitted packet, and the STATE 811 in which the file transmission / reception state and the congestion state of the TCP connection are changed is communicated. It transmits to the state change part 104.

  When the apparatus 200 receives the RST-ACK packet 1310 in which the TCP Flag 710 is RST-ACK from the server 504, the process of Step 979 is executed via Steps 921 and 923. In the process of step 979, the communication state is changed, that is, STATE 811 is changed to Close. Further, the operation matrix 207 whose mounting logic has been changed by the processing of step 1042 returns an RST-ACK packet 1311 in which the TCP Flag 710 is RST-ACK.

  As described above, the processing described with reference to FIG. 13 allows the apparatus 200 to download a file from the server using the file request packet from the access terminal 1200 as a trigger.

  FIG. 14 shows a sequence in which the device 200 receiving the GET command for requesting the file A using the application layer protocol HTTP transmitted from the access terminal 1200 divides and returns the file A stored in the device. The figure is shown. Here, the external memory 205 includes a pointer table 1409 and a file table 1410 in advance.

  When the device 200 receives the packet 1301 describing the GET command requesting the file A using the application layer protocol HTTP, the device 200 executes the processing of step 986 via steps 939, 941 and 944 of FIG. . In step 986, TSEQ806 is changed to PACK712, TACK807 is changed to the sum of PSE711, various commands 714 and Payload 715, and the communication state is changed, that is, STATE 811 is changed to HTTP GET Start. Further, the operation matrix 207 whose mounting logic has been changed by the processing of step 1047 in FIG. 10 searches the pointer table 1312 and determines whether or not the file A is stored in the external memory 205 (processing 1408). When the file A is accumulated, the TCP Flag 710 is ACK, and an ACK packet 1401 including the divided data 1411 of the file A is returned. Further, a memory 205 storing the changed congestion window size 808, transmitted window size SWIN 809, transmitted packet ID 812, STATE 811 in which the file transmission / reception state is changed to Active, and the end of the ACK received data are stored. The communication state 155 including the address pointer ADD813 is output to the communication state changing unit 104.

  When the apparatus 200 receives an ACK packet 1402/1404 whose TCP Flag 710 is ACK from the access terminal 1200, the apparatus 200 performs the process of Step 970, 971 or 972 via Step 958, 961, 964 or 967. In the process of step 970, 971 or 972, TSEQ806 is changed to PACK712, TACK807 is changed to the sum of PSEQ711, various commands 714, and Payload715, and the communication state is changed, that is, STATE811 is set to Slow Start / ConsistencyAvailance / FasterRevance Change to one of the following.

  Further, in the operation matrix 207 in which the mounting logic is changed by the processing of Steps 1044, 1045, or 1046 in FIG. 10, the TCP Flag 710 is ACK, and an ACK packet 1403 including the divided data 1411 of the file A is returned. Furthermore, the changed congestion window size 808, transmitted window size SWIN 809, transmitted packet ID 812, STATE 811 in which the file transmission / reception state and the TCP connection congestion state are changed, and the end of the ACK received data are stored. The communication state 155 including the address pointer ADD813 of the memory 205 is transmitted to the communication state changing unit 104. When the ACK packet 1404 is received, it indicates that the file A has already been transmitted. Therefore, the STATE 811 in the communication state 152 is changed to Passive indicating completion of the file transmission / reception state, and the packet is not returned.

  When apparatus 200 receives FIN-ACK packet 1405 in which TCP Flag 710 is FIN-ACK from access terminal 1200, the process of step 977 is executed via steps 915 and 917. In the process of step 977, the communication state is changed, that is, STATE 811 is changed to CLOSE, TACK 807 is changed to PSEQ 711 + 1, and TSEQ 806 is changed to PACK 712. Further, the operation matrix 207 whose mounting logic has been changed by the processing of Step 1041 returns a FIN-ACK packet 1406 in which the TCP Flag 710 is FIN-ACK. At this time, since the output STATE 811 in the communication state 155 is CLOSE, the apparatus 200 discards the ACK packet 1407 received thereafter by the process of step 973.

  As described above, the process described with reference to FIG. 14 allows the apparatus 200 to return a request file in response to the file request packet from the access terminal 1200.

  FIG. 15 shows a case where an apparatus 200 that has received a POST command transmitted by an access terminal 1200 and that has a SELECT command as an argument for requesting extraction of data that satisfies “TABLE1 1510 to I1 = A” using the application layer protocol HTTP. The sequence diagram which returns an extraction result is shown. It is assumed that the memory 205 has a database table TABLE1 1510 prepared therein.

  When the apparatus 200 receives a packet 1501 that describes a POST command having an argument of a SELECT command that requests extraction of data satisfying I1 = A from TABLE1 1510 using the application layer protocol HTTP, step 947 in FIG. The process of step 987 is executed via 949. In the process of step 987, TSEQ 806 is changed to PACK 712, TACK 807 is changed to the sum of PSE 711, various commands 714, and Pay load 715, and the communication state is changed, that is, STATE 811 is changed to HTTP POST Select Start.

  Further, the operation matrix 207 whose mounting logic has been changed by the processing of step 1049 in FIG. 10 searches the data 1511 satisfying I1 = A from TABLE1 1510, extracts this (processing 1509), and TCP Flag 710 is ACK. Yes, an ACK packet 1502 including the divided data of the extracted data 1511 is returned. Further, a memory 205 storing the changed congestion window size 808, transmitted window size SWIN 809, transmitted packet ID 812, STATE 811 in which the file transmission / reception state is changed to Active, and the end of the ACK received data are stored. The communication state 155 including the address pointer ADD813 is output to the communication state changing unit 104.

  When the device 200 receives an ACK packet 1503/1505 in which the TCP Flag 710 is ACK from the access terminal 1200, the device 200 performs the processing of Step 970, 971, or 972 via Step 958, 961, 964, or 967. In the process of step 970, 971 or 972, TSEQ806 is changed to PACK712, TACK807 is changed to the sum of PSEQ711, various commands 714, and Payload715, and the communication state is changed, that is, STATE811 is set to Slow Start / ConsistencyAvailance / FasterRevance Change to one of the following.

  Further, in the operation matrix 207 in which the mounting logic is changed by the processing of Steps 1044, 1045, or 1046 in FIG. 10, the TCP Flag 710 is ACK, and an ACK packet 1504 including the divided data of the extracted data 1511 is returned. Furthermore, the changed congestion window size 808, the transmitted window size SWIN 809, the ID number ID 812 of the transmitted packet, the STATE 811 in which the file transmission / reception state and the TCP connection congestion state are changed, and the end of the ACK received data are The communication state 155 including the stored address pointer ADD 813 of the memory 205 is output to the communication state changing unit 104. When the ACK packet 1505 is received, since all the extracted data 1511 has been transmitted, STATE 811 in the communication state 152 is changed to Passive indicating completion of the file transmission / reception state, and no packet is returned.

  When apparatus 200 receives FIN-ACK packet 1506 in which TCP Flag 710 is FIN-ACK from access terminal 1200, the process of step 977 is executed via steps 915 and 917. In step 977, the communication state is changed, that is, STATE 811 is changed to CLOSE, TACK 807 is changed to PSEQ 711 + 1, and TSEQ 806 is changed to PACK 712. Further, the operation matrix 207 whose mounting logic has been changed by the processing of Step 1041 returns a FIN-ACK packet 1507 in which the TCP Flag 710 is FIN-ACK. Further, since the STATE 811 of the communication state 155 that has been output is CLOSE, the ACK packet 1508 that the apparatus 200 receives thereafter is discarded by the process of step 973.

  As described above, the processing described with reference to FIG. 15 allows the apparatus 200 to return the selection result in response to the data selection request packet from the access terminal 1200.

  FIG. 16 shows an INSERT command that is transmitted from the access terminal 1200 and requests insertion of new data (I1, I2, I3) = (A3, B3, C3) to TABLE1 1612 using the application layer protocol HTTP. FIG. 10 shows a sequence diagram in which the apparatus 200 that receives the POST command having “” as an argument inserts new data into TABLE 1 1612 and uploads the insertion result to the server 504. In this example, the external memory 205 includes a TABLE1 1612 therein and a pointer table 1613 in which a plurality of pointers to data to be uploaded are described in advance.

  The apparatus 200 uses the application layer protocol HTTP to send a POST command having an INSERT command as an argument for requesting insertion of new data (I1, I2, I3) = (A3, B3, C3) to the TABLE1 1612. When the described packet 1601 is received, the process of step 989 is executed via steps 947, 949, 952 and 955 of FIG. In the process of Step 989, TSEQ 806 is changed to PACK 712, TACK 807 is changed to the sum of PSE 711, various commands 714, and Pay load 715, and the communication state is changed, that is, STATE 811 is changed to HTTP POST Insert Start.

  Furthermore, the operation matrix 207 whose mounting logic has been changed by the processing of step 1051 in FIG. 10 inserts new data (I1, I2, I3) = (A3, B3, C3) into TABLE1 1612 and pointers A pointer to data (I1, I2, I3) = (A3, B3, C3) is recorded in the table 1613 (process 1611). Further, when the operation matrix 207 determines that a certain amount of pointers are accumulated in the pointer table 1613 and the data to be uploaded exceeds a certain amount, the operation matrix 207 starts uploading the data to the server. At this time, a packet 1602 in which the TCP Flag 710 is set to SYN is generated and transmitted to the server 504. Further, a new communication state 155 between the apparatus 200 and the server 504 is output to the communication state changing unit 104.

  When the apparatus 200 receives the SYN-ACK packet 1603 in which the TCP Flag 710 is SYN-ACK from the server 504, the process of step 983 is performed via steps 933 and 935. In the process of step 983, TSEQ806 is changed to PACK712, and the communication state is changed, that is, STATE 811 is changed to FULL OPEN. Further, the operation matrix 207 whose mounting logic has been changed by the processing of step 1043 in FIG. 10 is the new data (I1, I2, I3) using the ACK packet 1604 in which the TCP Flag 710 is ACK and the application layer protocol HTTP. = A packet 1605 describing a POST command having an INSERT command as an argument for requesting insertion of (A1, B1, C1) to (A3, B3, C3) is transmitted. Data 1614 to be inserted is created using the pointer table 1613 and TABLE1 1612. Further, communication including the changed congestion window size 808, the ID 812 of the transmitted packet, the STATE 811 in which the file transmission / reception state is changed to Active, and the address pointer ADD813 of the memory 205 in which the end of the ACK received data is stored The state 155 is output to the communication state changing unit 104.

  When the device 200 receives the ACK packet 1606/1608 in which the TCP Flag 710 is ACK from the server 504, the device 200 executes the processing of Step 970, 971, or 972 via Step 958, 961, 964, or 967. In the process of step 970, 971 or 972, TSEQ806 is changed to PACK712, TACK807 is changed to the sum of PSEQ711, various commands 714, and Payload715, and the communication state is changed, that is, STATE811 is set to Slow Start / ConsistencyAvailance / FasterRevance Change to one of the following.

  Further, in the operation matrix 207 in which the mounting logic is changed by the processing of Steps 1044, 1045 or 1046 in FIG. 10, the TCP Flag 710 is ACK, and an ACK packet 1607 in which the continuation of the packet 1605 is described is returned. Further, the communication state 155 including the changed congestion window size 808, the transmitted window size 809, the ID number ID 812 of the transmitted packet, and the STATE 811 in which the file transmission / reception state and the congestion state of the TCP connection are changed is communicated. The data is output to the state change unit 104.

  In addition, when the operation matrix 207 receives the ACK packet 1608, since all the continuation of the POST command that could not be described in the packet 1605 has been transmitted, STATE 811 is changed to Passive indicating completion of the file transmission / reception state, Packets are not returned. Further, all the contents of the pointer table 1613 are erased.

  When the apparatus 200 receives the RST-ACK packet 1609 in which the TCP Flag 710 is RST-ACK from the server 504, the process of step 979 is executed via steps 921 and 923. In the process of step 979, the communication state is changed, that is, STATE 811 is changed to Close). Further, the operation matrix 207 whose mounting logic has been changed by the processing of step 1042 returns an RST-ACK packet 1610 in which the TCP Flag 710 is RST-ACK.

  As described above, the processing described with reference to FIG. 16 allows the apparatus 200 to update the database table in response to the data insertion request packet from the access terminal 1200 and upload the file to the server.

  FIG. 17 shows a case where the apparatus 200 that has received a POST command having an argument of a DELETE command that requests deletion of data satisfying I1 = A from TABLE1 1710 using the application layer protocol HTTP transmitted from the access terminal 1200 is deleted. The sequence diagram which returns a result is shown. It is assumed that the memory 205 has a database table TABLE1 1710 prepared therein.

  When the apparatus 200 receives a packet 1701 describing a POST command having a DELETE command as an argument for requesting deletion of data satisfying I1 = A from TABLE1 1710 using the application layer protocol HTTP, step 947 in FIG. The process of step 988 is executed via 949 and 952. In the process of step 988, TSEQ 806 is changed to PACK 712, TACK 807 is changed to the sum of PSE 711, various commands 714, and Pay load 715, and the communication state is changed, that is, STATE 811 is changed to HTTP POST Delete Start.

  Further, the operation matrix 207 whose mounting logic has been changed by the processing of step 1050 in FIG. 10 deletes data satisfying I1 = A from TABLE1 1710 (processing 1709), TCP Flag 710 is ACK, and the deletion result 1711 An ACK packet 1702 including the divided data is returned. Furthermore, the changed congestion window size 808, transmitted window size SWIN 809, transmitted packet ID 812, STATE 811 in which the file transmission / reception state has changed to Active, and the end of the ACK received data are stored in the memory 205. The communication state 155 including the address pointer ADD813 is output to the communication state changing unit 104.

  When the apparatus 200 receives an ACK packet 1703/1705 in which the TCP Flag 710 is ACK from the access terminal 1200, the apparatus 200 performs the process of Step 970, 971, or 972 via Step 958, 961, 964, or 967. In the process of step 970, 971 or 972, TSEQ806 is changed to PACK712, TACK807 is changed to the sum of PSEQ711, various commands 714, and Payload715, and the communication state is changed, that is, STATE811 is set to Slow Start / ConsistencyAvailance / FasterRevance Change to Further, in the operation matrix 207 in which the mounting logic is changed by the processing of Steps 1044, 1045, or 1046, the TCP Flag 710 is ACK, and an ACK packet 1704 including the divided data of the deletion result 1711 is returned. Furthermore, the changed congestion window size 808, the transmitted window size SWIN 809, the ID number ID 812 of the transmitted packet, the STATE 811 in which the file transmission / reception state and the TCP connection congestion state have changed, and the end of the ACK received data The communication state 155 including the address pointer ADD813 of the memory 205 in which is stored is output to the communication state changing unit 104. When the ACK packet 1705 is received, the deletion result 1711 has already been transmitted, so the STATE 811 in the communication state 152 is Passive indicating completion of the file transmission / reception state, and the packet is not returned.

  When the apparatus 200 receives the FIN-ACK packet 1706 in which the TCP Flag 710 is FIN-ACK from the access terminal 1200, the process of step 977 is executed via steps 915 and 917. In the process of step 977, the communication state is changed, that is, STATE 811 is changed to CLOSE, TACK 807 is changed to PSEQ 711 + 1, and TSEQ 806 is changed to PACK 712. Further, the operation matrix 207 whose mounting logic has been changed by the processing of Step 1041 returns a FIN-ACK packet 1707 in which the TCP Flag 710 is FIN-ACK. At this time, since the output STATE 811 in the communication state 155 is CLOSE, the ACK packet 1708 received by the apparatus 200 thereafter is discarded by the process of step 973.

  As described above, the apparatus 200 can delete the data from the database table and return the deletion result in response to the data deletion request packet from the access terminal 1200 by the processing described with reference to FIG.

  As described above, the processing described with reference to FIGS. 11 to 17 enables the apparatus 200 according to the embodiment of the present invention to execute a server service of an application layer such as a database based on a request from an access terminal or the like. Become.

1 is a configuration block diagram of an apparatus according to an embodiment of the present invention. It is a block diagram of the configuration of the dynamic reconfiguration processor of the embodiment of the present invention. It is explanatory drawing of operation | movement of the dynamic reconfiguration processor of embodiment of this invention. It is explanatory drawing of the system of embodiment of this invention. It is explanatory drawing of the system of embodiment of this invention. FIG. 2 is a configuration block diagram of an external controller, a dynamic reconfiguration processor, and a memory according to the embodiment of this invention. It is explanatory drawing of the packet data of embodiment of this invention. It is explanatory drawing of an example of the communication status table of embodiment of this invention. It is explanatory drawing of an example of the combination of the communication state of embodiment of this invention. It is a flowchart of a process of the communication state change part of embodiment of this invention. It is a flowchart of a process of the communication status classification | category part of embodiment of this invention, and the arithmetic unit for a reconstruction trigger production | generation. It is a sequence diagram which shows the case where the abnormal communication of embodiment of this invention is excluded. It is a sequence diagram which shows the case where normal communication of embodiment of this invention is protected. It is a sequence diagram of the process which downloads the file from the server of embodiment of this invention. It is a sequence diagram which returns a request file with respect to the file request packet of embodiment of this invention. It is a sequence diagram of the process which returns a selection result with respect to the data selection request packet of embodiment of this invention. It is a sequence diagram of the process which updates a database table with respect to the data insertion request packet of embodiment of this invention, and uploads the file to a server. It is a sequence diagram of the process which deletes data from a database table with respect to the data deletion request packet of embodiment of this invention, and returns a deletion result. It is explanatory drawing of operation | movement of the conventional general purpose processor. It is explanatory drawing of the operation | movement of the conventional communication data input / output of a processor.

Explanation of symbols

200 Device 202 External Controller 201 Dynamic Reconfiguration Processor 205 Memory 501 Access Network 502 Edge Network 503 Back-end System

Claims (9)

  An information processing apparatus that is connected to a network, receives a packet transferred through the network, and executes processing according to a communication state between hosts that transmit and receive the packet,
  An interface for transmitting and receiving the packet;
  A transition / congestion state of a transport layer protocol between the hosts, a type of an application layer protocol, a command being executed by the application layer protocol and a type of an argument of the command, and a progress state of execution of the command, A communication state storage unit that stores a communication state including a communication state table;
  Connected to the interface and the communication state storage unit, reads the communication state from the communication state storage unit, changes the read communication state, and writes the changed communication state to the communication state storage unit , A communication state changing unit that outputs the changed communication state;
  An arithmetic unit having a plurality of functions connected to the communication state changing unit is provided, the mounted logic is changed by changing the wiring of the arithmetic unit group, and the communication state is output by the communication state changing unit. An arithmetic matrix for processing the received packet based on:
  A circuit configuration accumulation buffer that is connected to the arithmetic matrix and stores configuration information for specifying the function of the arithmetic unit and mutual wiring;
  The communication state changing unit
  Based on the information included in the packet received by the interface, the communication state between the hosts that transmit and receive the packet is read from the communication state storage unit,
  Based on the information included in the received packet and the content of the read communication state, the read communication state is changed, and the changed communication state is written in the communication state storage unit. , Outputting the changed communication state to the operation matrix,
  The arithmetic matrix is
  Based on the communication state output from the communication state change unit, to obtain one configuration information from the circuit configuration accumulation buffer,
  Based on the acquired configuration information, the function of the arithmetic unit group and the mutual wiring are reconfigured, the mounting logic in the arithmetic matrix is changed, and the received packet corresponds to the communication state. An information processing apparatus that executes processing.   The communication state storage unit
  The communication state between the hosts is stored as one entry for each set of a source IP address, a destination IP address, a TCP source port number, and a TCP destination port number,
  The entry includes a value indicating the start or stop of the TCP connection, a value indicating the establishment or non-establishment of the TCP connection, a value indicating the congestion state of the TCP connection, the sequence number of the sending host of the TCP connection, and the TCP connection. The receiving host's sequence number, the congestion window size indicating the data size that can be transmitted at one time, the transmitted window size indicating the transmitted data size, the latest time when the packet was received, and the ID number of the transmitted packet , The address pointer of the memory storing the end of the ACK received data, the value indicating the presence / absence of the protocol used by the application layer and the type of the protocol, the presence / absence of the command being executed by the protocol and the command argument A value indicating the type and the protocol Start to send and receive files running the command, stores a value indicating the incomplete or complete, and
  The communication state changing unit
  Based on the source IP address, destination IP address, TCP source port number, and TCP destination port number included in the header of the packet received by the interface, between the hosts that transmit and receive the packet The information processing apparatus according to claim 1, wherein the communication state entry is read from the communication state storage unit.   A packet buffer connected to the interface and the operation matrix, temporarily storing the received packets and then outputting the packet to the operation matrix;
  A communication state storage buffer connected to the communication state changing unit and the calculation matrix, and temporarily storing the communication state output by the communication state changing unit and then outputting to the calculation matrix;
  The arithmetic matrix is
  For each acquired packet,
  Obtaining a packet output from the packet buffer and a communication state output from the communication state storage buffer;
  Based on the acquired configuration information, the function of the arithmetic unit group and the mutual wiring are reconfigured, the mounting logic in the arithmetic matrix is changed, and the received packet corresponds to the communication state. The information processing apparatus according to claim 1, wherein processing is executed.   A packet determination unit connected to the calculation matrix, classifying the packet based on information included in the packet, and outputting a classification result of the classified packet to the calculation matrix;
  The communication state determination unit, the packet determination unit, the communication state change unit, and the calculation matrix are connected to the classification result output by the communication state determination unit and the classification result output by the packet determination unit. A delay adjusting unit that adjusts a delay generated in the communication state with the packet, and outputs the packet and the communication state to the operation matrix according to the adjusted delay,
  The arithmetic matrix is
  One configuration information corresponding to the classification result of the communication state output by the communication state determination unit and the classification result of the packet output by the packet determination unit is acquired from the circuit configuration accumulation buffer, and the acquired Reconfigure the functions of the computing unit group and the mutual wiring based on the configuration information, and change the mounting logic of the computation matrix,
  The delay adjustment unit
  Based on the classification result of the communication state output by the communication state determination unit and the classification result of the packet output by the packet determination unit, predict the timing at which the reconstruction of the operation matrix ends, or Received the notification of the completion of reconstruction from the calculation matrix,
  The information processing apparatus according to claim 1, wherein a delay to be generated between the packet and the communication state is changed based on the predicted timing or the notification of completion of the reconfiguration.   The communication state changing unit includes a combination of a TCP flag number included in a TCP header of the received packet and a value indicating establishment or non-establishment of a TCP connection included in the communication state, and a TCP header of the received packet. A combination of a transmission or reception sequence number included in the communication state and a transmission or reception sequence number included in the communication state, and a TCP flag number included in a TCP header of the received packet and an application layer protocol command included in the communication state The information processing apparatus according to claim 1, wherein the communication state is changed based on at least one combination of a value indicating completion or incompleteness of file transmission / reception being executed.   A communication state determination unit that is connected to the communication state change unit and the calculation matrix, classifies the communication state output by the communication state change unit, and outputs a classification result of the classified communication state;
  Connected to the communication state determination unit, the communication state change unit, and the calculation matrix, and based on the classification result output by the communication state determination unit, adjust the delay generated in the packet and the communication state, A delay adjustment unit that outputs the packet and the communication state to the arithmetic matrix by the adjusted delay, and
  The arithmetic matrix is
  One configuration information corresponding to the communication status classification result output by the communication status determination unit is acquired from the circuit configuration storage buffer, and the function of the computing unit group and the mutual wiring based on the acquired configuration information And change the mounting logic of the operation matrix,
  The delay adjustment unit
  Based on the classification result of the communication state output by the communication state determination unit, predict the timing when the reconstruction of the operation matrix ends, or receive the notification of the completion of the reconstruction output from the operation matrix,
  The information processing apparatus according to claim 1, wherein a delay to be generated between the packet and the communication state is changed based on the predicted timing or the notification of completion of the reconfiguration.   The calculation matrix generates an autonomous reconfiguration trigger for requesting acquisition and reconfiguration of configuration information stored in the circuit configuration storage buffer based on the classification result output by the communication state determination unit. Computation group for generating configuration triggers
  The configuration information is selected from the circuit configuration accumulation buffer based on the autonomous reconfiguration trigger generated by the autonomous reconfiguration trigger generation computing unit connected to the operation matrix and the circuit configuration accumulation buffer, and the selection An autonomous reconfiguration mechanism that outputs the configured information to the calculation matrix,
  The arithmetic matrix is
  Based on the configuration information output by the autonomous reconfiguration mechanism, reconfigure the functions of the arithmetic unit group and the mutual wiring, change the mounting logic inside the arithmetic matrix, and change the received packet to the received packet. The information processing apparatus according to claim 6, wherein a process corresponding to a communication state is executed.   An external memory that is connected to the arithmetic matrix and includes a circuit configuration storage area that stores a plurality of the configuration information, and a pointer area that stores a pointer for each configuration information;
  The calculation matrix generates an external reconfiguration trigger for requesting acquisition and reconfiguration of configuration information stored in the external memory based on the classification result output by the communication state determination unit, and the external memory Comprising an arithmetic unit group for generating an external reconfiguration trigger for updating a pointer stored in the pointer area of
  A pointer is connected to the external memory, the arithmetic matrix, and the circuit configuration accumulation buffer, and a pointer is acquired from a pointer area of the external memory based on an external reconfiguration trigger generated by the external reconfiguration trigger generation arithmetic unit group. One configuration information is acquired from the circuit configuration storage area of the external memory based on the pointer, the acquired configuration information is written to the circuit configuration storage buffer, and the written configuration information is output to the arithmetic matrix. A processor for executing the circuit configuration change module
  The arithmetic matrix is
  Based on the configuration information output by the circuit configuration change module, the function of the arithmetic unit group and the mutual wiring are reconfigured, the mounted logic inside the arithmetic matrix is changed, and the received packet is changed to the received packet. The information processing apparatus according to claim 6, wherein a process corresponding to a communication state is executed.   The communication state determination unit
  Classifying the communication state by determining the start or stop of the TCP connection with reference to a value indicating the start or stop of the TCP connection included in the communication state output by the communication state change unit;
  If the TCP connection is started, refer to the value indicating the establishment or non-establishment of the TCP connection included in the communication state, classify the communication state by determining the establishment or non-establishment of the TCP connection,
  If a TCP connection is established, refer to a value indicating the congestion state of the TCP connection included in the communication state, classify the communication state by determining the congestion state of the TCP connection,
  Classifying the communication state by determining the presence or absence of an application layer protocol included in the communication state;
  If there is an application layer protocol, classify the communication state by determining the type of application layer protocol included in the communication state, the command being executed by the protocol, and the presence of arguments of the command,
  If there is an argument of the command and the command, determine the type of the command and the argument of the command, and determine whether the progress status of the file transmission / reception being executed by the command is started, incomplete or completed The information processing apparatus according to claim 6, wherein the communication state is classified by doing so.