JP6499567B2 - Virtual machine placement destination determination apparatus and virtual machine placement destination determination method - Google Patents

Description

  The present invention relates to a technology for a virtual machine placement destination determination apparatus and a virtual machine placement destination determination method.

2. Description of the Related Art A virtualization system that efficiently uses computer resources of a physical server by placing and executing a virtual machine (VM) on the physical server has become widespread.
Japanese Patent Application Laid-Open No. 2004-228561 describes that a physical server load is predicted from application processing schedule information, and is arranged in a combination of virtual machines with a small deviation between the resource consumption amounts of the physical server.
Patent Document 2 describes an arrangement method in which virtual machine accommodation efficiency is improved while ensuring availability while taking into account the types of active and standby systems of virtual machines.

JP 2013-239095 A JP 2015-22671 A

  As the use of cloud services expands, an increasing number of cloud service providers manage DC bases distributed in multiple locations as bases (hereinafter referred to as DC bases) of data centers (DCs) that are deployment destinations. . In addition, scenes in which the same cloud service is used across a plurality of user bases (branches, sales offices, etc.) in major companies and the like are increasing. As the number of bases increases, the bases are distributed in various places. Therefore, the communication path from the user base that is the virtual machine usage source to the DC base that is the placement destination of the virtual machine tends to become a long distance year by year, such as across prefectures and countries.

  Since cloud services provide services via a network, the communication performance of the network is often affected by the quality of the service. For example, in a cloud service in which two-way conversation is performed, if the communication time is long, the delay reduces the satisfaction of the user. Therefore, it is necessary not only to guarantee the connection from geographically distributed user bases to one cloud service, but also to guarantee the communication quality (required communication requirement) of the connection.

FIG. 9 is an explanatory diagram showing a breakdown of service response times.
A period from when the client terminal in the user base receives the processing request until the response is output is referred to as a “response period”. This response period is the sum of the following periods (time).
・ Period for accepting input of processing request ・ Uplink communication time for sending processing request to DC base ・ Period for system processing in DC base that received processing request ・ Downlink communication time for receiving system processing result from DC base ・Period during which received system processing results are output as a response

  In order to shorten the response time, it is effective to shorten the uplink and downlink communication times by arranging the user base and the DC base close to each other. Movement of the user base is difficult due to the convenience of the user. On the other hand, the communication time varies depending on which DC base is arranged with the cloud service among a plurality of DC bases managed by the same cloud service provider. That is, there is a possibility that the communication time can be shortened by reducing the number of communication hops of the communication path by determining the arrangement destination in a wide range of data center level.

  However, in the conventional virtual machine placement technologies such as Patent Documents 1 and 2, the performance of the physical server itself that is the placement destination (computer resources such as CPU resources and memory capacity) is considered, but the communication path is long distance As a result, it was not assumed that the communication quality would vary. Therefore, even if there is room in the internal processing of the physical server, the user's satisfaction may not be satisfied due to the lack of communication performance when transmitting the result.

  Accordingly, the main object of the present invention is to arrange virtual machines used from a plurality of user bases at different positions at positions where appropriate communication quality can be provided.

In order to solve the above-described problem, the virtual machine placement destination determination device of the present invention includes:
Position information on the network topology for each placement base that includes physical servers that are candidates for placement of virtual machines, and position information on the network topology for each user base that includes a user terminal that uses the virtual machine; A storage unit storing connection information of each communication device constituting the network topology;
By referring to the connection information of each communication device, the communication performance when communicating from each user base to each placement base is specified, and the communication performance satisfying the predetermined communication requirement from any user base the arrangement bases capable of providing, possess a placement destination determining section for determining a placement destination of the virtual machine,
The placement destination determination unit
When there is no placement base that can provide communication performance that satisfies the predetermined communication requirements from any of the user bases, communication performance that can be provided to the user base with a high priority set in advance is higher than other placement bases. Determine the location where the virtual machine is to be placed as the placement destination of the virtual machine,
When there are a plurality of users having one or more user bases, the order of users who arrange the virtual machines is determined according to the following rules.
(Rule 1) Users who require higher communication performance as the predetermined communication requirement are selected first.
(Rule 2) When a tie is found in the above (Rule 1), a user with more user bases is selected first.
(Rule 3) When the score is the same in the above (Rule 2), the user having a higher priority user base is selected first.

As a result, the communication performance from the user base to the placement base is reflected in the determination process of the placement base of the placement destination of the virtual machine, so that the virtual machines used from the user bases at a plurality of different positions can have an appropriate communication quality. Can be provided at a position where the
Furthermore, it is possible to suppress the user's satisfaction from being greatly reduced even in a situation where communication resources tend to be insufficient.
Then, by arranging the virtual machines of users who are difficult to arrange first, the number of users satisfying the communication requirements can be increased.

  In the present invention, the placement destination determination unit refers to the connection information of each communication device, specifies the number of hops from the communication device housing each user base to the communication device housing each placement base, The communication time proportional to the specified number of hops is specified as communication performance from each user base to each placement base.

  Thereby, the communication time can be easily calculated from the number of hops.

  In the present invention, when the placement base determined as the placement destination of the predetermined virtual machine by the placement destination determination unit cannot provide the computer resource required by the predetermined virtual machine, the placement destination determination unit is already placed on the determined placement base. After moving another virtual machine to another placement base, the predetermined virtual machine is placed at the determined placement base.

  Thereby, the number of users satisfying the communication requirements can be increased by making the arrangement bases interchangeable among users.

  ADVANTAGE OF THE INVENTION According to this invention, the virtual machine utilized from the user base of several different positions can be arrange | positioned in the position which can provide suitable communication quality.

It is a block diagram of the network for cloud service construction concerning this embodiment. It is a block diagram of the arrangement | positioning destination determination apparatus concerning this embodiment. FIG. 3A is a configuration diagram of the user base information. FIG. 3B is a configuration diagram of DC base information. FIG. 3C is a configuration diagram of the arrangement service information. FIG. 4A is a diagram illustrating a hierarchical structure of a cloud service construction network. FIG. 4B is a configuration diagram of communication facility information. FIG. 5A is a table showing combinations of communication costs between bases. FIG. 5B is a table showing whether or not the communication cost between bases satisfies the allowable time. FIG. 5C is a table for selecting a destination DC base based on the communication cost between bases. It is a flowchart which shows the processing content of the arrangement | positioning destination determination part concerning this embodiment. It is a figure which shows the arrangement | positioning condition of the some user concerning this embodiment. It is the figure which performed the two-stage arrangement | positioning of the service about the network for cloud service construction of FIG. 1 concerning this embodiment. It is explanatory drawing which shows the breakdown of the response time of the service concerning this embodiment.

  Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a configuration diagram of a cloud service construction network. In this network system, routers R (R11 to R15, R21, R31, R51), which are communication devices, are connected (thick lines in the figure). Further, each router has a plurality of user bases U (UA1 to UA4) used by the same user UA and a DC serving as a placement base that is a placement destination of a virtual machine (VM) for executing a cloud service. Base D (D1-D4) is connected. Note that the cloud service is processing content described in a program or the like, and functions by being executed from a virtual machine arranged in a physical server.

  One or more terminals in a user base (for example, UA1) access a cloud service (for example, employee search DB) executed by a VM built on a physical server in a certain DC base (for example, D3). Communication (for example, 4 hops of R21 → R12 → R13 → R14). On the other hand, by arranging the employee search DB at the DC site D1, the communication distance from the user site UA1 can be shortened (2 hops from R21 to R12), and the communication time from the DC site D1 can be shortened. .

  However, the employee search DB on the DC site D3 is used from all user sites (UA1 to UA4) of the same user UA. The DC site D1 is a short distance from the user site UA1, but is a long distance from the user site UA3 (4 hops of R51 → R15 → R11 → R12). In other words, the cloud service provided in the present embodiment has a DC site D located at a communication distance that can be accommodated within a communication time within a range of allowable communication time from any user site U belonging to the same user as an arrangement destination. Must be selected.

FIG. 2 is a configuration diagram of the placement destination determination apparatus 1.
The placement destination determination device 1 is configured as a computer having a CPU (Central Processing Unit), a memory, storage means (storage unit) such as a hard disk, and a network interface.
In this computer, the CPU executes a program (also referred to as an application or its abbreviated application) read on the memory, thereby operating a control unit (control means) configured by each processing unit described later.

  The placement destination determination device 1 is a server that manages the cloud service construction network of FIG. 1 and determines the DC site D that is the placement destination of the cloud service (VM for executing). Therefore, the placement destination determination apparatus 1 includes a user information input unit 10, a DC information input unit 20, a placement destination determination unit 31, and a placement instruction unit 32 as processing units. Furthermore, the placement destination determination device 1 transmits user base information 11, placement service information 12, DC base information 21, and communication facility information 22 as data for the processing units to execute processing inside or outside the device. Store in the storage means (storage unit). Details of each of these data will be described later with reference to FIGS.

The user information input unit 10 causes the user to input user base information 11 and arrangement service information 12 as information related to the user base U of the network for building a cloud service.
The DC information input unit 20 causes a service provider to input DC base information 21 and communication facility information 22 as information about the DC base D and the router R of the cloud service construction network.
The placement destination determination unit 31 determines the DC base D of the cloud service placement destination that guarantees the allowable communication time based on the information received by the user information input unit 10 and the DC information input unit 20. .
The placement instruction unit 32 instructs the virtual infrastructure system in the DC destination D of the placement destination determined by the placement destination determination unit 31 to place a virtual machine for executing the cloud service. The virtual infrastructure system in the DC site D receives an instruction from the arrangement instruction unit 32 and arranges the corresponding virtual machine.

FIG. 3A is a configuration diagram of the user base information 11.
For each user name, the user base information 11 includes a list of user bases from which the user uses the cloud service, a nearest router (for example, one hop) R indicating the location of the user base, The priority of the user base (high / normal) is associated. Each user base may be distributed in various parts of Japan such as Tokyo, Aomori, Yamanashi, Okinawa, etc., and as shown in FIG. As a result, the communication time may vary.

FIG. 3B is a configuration diagram of the DC base information 21.
In each DC site, the DC site information 21 associates the current available resource, the nearest router R indicating the location of the DC site, and the availability.
The “free resource” column of the DC site information 21 is a total for each site regarding the resources that the physical server in the DC site can currently provide. This “free resource” column is described for each type of computer resource. In FIG. 3B, as an example, the number of CPU cores, memory capacity, disk I / O speed, and NW (NetWork) I / O speed are illustrated. Has been.

  As a resource amount notation for each resource type, a plurality of types of resources may be collectively expressed as one virtual machine performance (for 1 VM). For example, the computer performance used per unit VM is 8 cores for CPU, 10 GB for memory, 1 Gbps for disk I / O, and 0.5 Gbps for network I / O. Here, the DC base D1 can provide 80 cores of CPU, 100 GB of memory, 10 Gbps of disk I / O, and 5 Gbps of network I / O, so the unit VM is 10 units (10 VM) is there.

  The “vacancy status” column of the DC site information 21 is a column indicating a determination result as to whether or not each cloud service can be arranged in the DC site. FIG. 3B illustrates a determination result when paying attention to the employee search DB (see FIG. 3C).

FIG. 3C is a configuration diagram of the arrangement service information 12.
The arrangement service information 12 associates the necessary resources with the necessary communication requirements for providing the service for each service name of the cloud service. The required resource has the same notation as the “free resource” column described with reference to FIG. 3B, and is the consumption amount of the resource of the physical server in the DC site D accompanying the execution of the cloud service.
If (the resource amount in the “necessary resource” column in FIG. 3C) <(the resource amount in the “free resource” column in FIG. 3B), the “free status” column = “present” in FIG. 3B. It is.

  The required communication requirement in FIG. 3C indicates a standard required as communication performance between the user base U and the DC base D accompanying execution of the cloud service. For example, the employee search DB is required to have a one-way communication time of user base U → DC base D or one way from user base U ← DC base D to 0.8 seconds or less. Furthermore, when defining the required communication requirements using an index of communication time, not only one-way communication time but also round-trip communication time and uplink and downlink communication times may be designated.

The necessary communication requirements vary depending on the type of service and the user of the service partner. For example, in a cloud gaming service that continues to send the game screen generated by the server to the client, it is necessary to immediately reflect the user's operation on the game screen, so the required communication requirements are strict (allowable communication time is short) Become.
In this embodiment, attention is paid to communication time (delay) as an example of necessary communication requirements. However, this is merely an example of a communication index, and other communication such as communication speed (bandwidth) and jitter (communication fluctuation) is used. The indicator may be used in combination with the communication time or alone. For example, in the video distribution system, the necessary communication requirement “bandwidth = 10 [Mbps] or more” is registered.

  Although not shown, additional information such as a computer environment in which the cloud service is executed may be stored in the arrangement service information 12 in association with the service name as information about the cloud service. The computer environment is, for example, an operating system to be executed (Ubuntsu 12.0 or the like), a Web server to be used (Apache 10.0 or the like), and a database server to be used (Oracle DB 5.0 or the like).

FIG. 4A is a diagram illustrating a hierarchical structure of a cloud service construction network.
Each router R is connected in an upper and lower hierarchical structure. For example, in FIG. 4A, the first layer (R1) of one group, the second layer (R11, R12, R13, R14, R15) of one group, and the third layer (“R21” of three groups). , R22, R23 ”group,“ R31, R32, R33 ”group,“ R51, R52 ”group).
In general, the physical distance between the routers R in the upper layer becomes longer. For example, the first layer connects between countries, the second layer connects between prefectures and prefectures, and the third layer connects between cities.

Communication from a router (for example, router R11) to another router (for example, router R13) in the same group is transmitted between adjacent routers via the ring network in the group (R11 → R15 → R14 → R13 if clockwise). ).
Communication from a router (for example, router R11) to another group of routers (for example, router R33) is the same as communication to another router of the same group described above, up to a connection router (for example, router R13) between groups. . Further, after being transferred from the connecting router R13 to the router R31 of another group, it is transmitted within another group (R31 → R32 → R33 if clockwise).

  Note that the network in FIG. 4A may be provided with redundancy so that communication can be performed bidirectionally between the active system (clockwise) and the standby system (counterclockwise). For example, in the normal state, the clockwise route “R11 → R15 → R14 → R13” is used. However, when a failure occurs in R14, the counterclockwise route “R11 → R12 → R13” is used. Good.

  FIG. 4B is a configuration diagram of the communication facility information 22. Here, the communication facility information 22 showing the cloud service construction network of FIG. The communication facility information 22 associates, for each router name, neighboring routers in the same group in the clockwise and counterclockwise direction of the router and routers directly connected to the upper and lower levels as viewed from the router indicated by the router name. ing. Although not shown, the communication facility information 22 also holds route policy information such as “the current system is clockwise and the standby system is counterclockwise” as described above.

  By referring to the network topology information of the communication facility information 22, for example, the communication performance (how many hops and how long the communication time is) from R11 to R33 can be specified. In order to specify the communication performance between routers in more detail, not only the connection information between routers (the adjacent router is R12, etc.) but also the communication performance related to the connection (the communication line speed to the adjacent router R12 is 100 [Mbps], the communication delay is 10 [ms], the average transfer processing time in the router R12 is 100 [ms], etc.) may be managed together with the communication facility information 22.

FIG. 5A is a table showing combinations of communication costs between bases. This table shows a calculation process during the processing of the placement destination determination unit 31.
First, the arrangement destination determination unit 31 refers to each piece of information such as the communication facility information 22 described with reference to FIG. 4B when there are four DC base candidates (D1 to D4) where a certain cloud service is arranged. Then, for each user base of the cloud service, the communication performance (how many hops and how long the communication time is) to each candidate DC base is obtained.

For example, consider the communication performance of the combination of user base UA1 → DC base D1 in the first column and first row of FIG. The placement destination determination unit 31 refers to the user base information 11 and acquires that the user base UA1 is closest to R21 and the DC base information 21 is referred to as R12 is the closest to the DC base D1.
Next, the placement destination determination unit 31 refers to the communication facility information 22 as the communication performance of R21 → R12, and it is found that the routers are directly connected as upper and lower levels. That is, it acquires that it passes through two routers R21 and R12 (two hops). Furthermore, for the sake of simplicity, assuming that processing time of 0.2 [s] per hop is required for all routers, the placement destination determination unit 31 sets the communication performance (communication time) of R21 → R12 to 2 [hops]. ] × 0.2 [s] = 0.4 [s]. When it is assumed that the time for transmitting light between 1 hop is as small as 0, 0.2 [s] is mainly determined by the processing time and communication delay in the router.
The communication time estimation method based on the number of hops can be easily calculated.

On the other hand, instead of the communication time estimation method based on the number of hops, the following communication time calculation method for each communication facility may be used.
Communication time of 2 sites (user site and DC site) = (total of internal processing time required for transfer processing of each communication device passing between the two sites) + (in the cable between each communication device passing between the two sites) Total transmission time)
The transmission time in the cable is determined by the total cable distance divided by the speed of light. However, in a cable laid in Japan, the total cable distance is short with respect to the speed of light. Good.

FIG. 5B is a table showing whether or not the communication cost between bases satisfies the allowable time. Here, the allowable time is 0.8 [s]. For each communication time in FIG. 5A, a communication time of 0.8 [s] or less is allowed (O), and a communication time exceeding 0.8 [s] is rejected (X).
The placement destination determination unit 31 selects, as the placement destination, the DC base D3 in which the communication time from all the user bases UA1 to UA4 is permissible (O). As a result, access is performed with equal communication quality from all the user bases whose positions are dispersed, so that, for example, services can be provided fairly even between user bases paying the same price.

FIG. 5C is a table for selecting a DC base as a placement destination based on the communication cost between bases, as in FIG. 5B. Here, since the allowable time = 0.6 [s] (because the restriction becomes tight), the number of allowed combinations (◯) is smaller than that in FIG. 5B. Therefore, there is no DC base where the communication time from all the user bases UA1 to UA4 is permissible (O).
Therefore, the placement destination determination unit 31 selects a DC base that can further reduce the sum (or average time) of communication times from the user bases UA2 and UA3 that have high priority as a next best measure. In other words, the placement destination determination unit 31 narrows down the user bases with high priority set in advance, and uses a DC base whose communication performance that can be provided to the user base is better than other DC bases as the placement destination of the virtual machine. decide.
However, in FIG. 5 (c), although the DC base D2 whose communication time sum is 1.6 [s] can be excluded, the remaining DC bases D1, D3, and D4 whose communication time sum is 1.4 [s] are the same. As a result, it is not possible to determine one DC base as the placement destination.

Furthermore, when it is not possible to determine one DC site as the placement destination, the placement destination determination unit 31 refers to the priority for each user base registered in the user base information 11 in advance, so that the user bases with high priority are not connected. A DC base with a small communication time difference (distance) is set as the placement destination. Therefore, as shown in the fifth row of FIG. 5C (the “high difference” indicating the communication time difference between the user bases UA having high priority), the placement destination determining unit 31 has the user base UA2 having high priority. , The difference in communication time between UA3 is calculated for each DC base. For example, at the DC site D1, (UA3 communication time 1.0 [s]) − (UA2 communication time 0.4 [s]) = 0.6 [s].
Then, the placement destination determination unit 31 sets the DC base D3 having the minimum difference time of 0.2 [s] among the difference times in the “high difference” row as the placement destination. As a result, in the high priority user bases UA2 and UA3 that are particularly emphasized by the user, it is possible to reduce the number of user bases with relatively poor quality when viewed relatively.
In the above example, there are two high-priority bases, so the difference between the bases is shown. However, in the case of three or more bases, an indicator showing the deviation such as the standard deviation of the communication time difference between the bases A value may be used.

  In the video conference distribution service, a large amount of video data collected at the site is distributed in real time from the distributor to the system located in the DC. Therefore, for example, the user base information 11 is registered in advance so that the high priority user base is the distributor (speaker captured in the video) and the normal user base is the receiver (listener who is not captured in the video). Keep it. As a result, by reducing the communication time delay during the meeting, it is possible to prevent the distributor and the listener from degrading the level of service satisfaction.

  Here, even between the same routers such as R11 → R13, the communication time of the active system such as “R11 → R15 → R14 → R13” in the clockwise direction (0.8 [s] in 4 hops) and the counterclockwise “ The communication time of the standby system such as “R11 → R12 → R13” (0.6 [s] for 3 hops) may be different. Therefore, when using a network with redundancy, the communication time of the active system may be used as it is as the communication time between bases, or the statistical time (minimum) of the communication time of the active system and the communication time of the standby system Time, maximum time, average time, total time, etc.) may be used.

Furthermore, as exemplified below, the both systems may be combined to perform the allowance (o) / rejection (x) determination with respect to the allowable time.
(Judgment example 1) Since the active system communication time (= 0.8 [s]) ≤ allowable time (= 0.8 [s]), allowable (○)
(Judgment example 2) Active system communication time (= 0.8 [s]) ≦ allowable time (= 0.8 [s]) and standby system communication time (= 0.6 [s]) ≦ allowable time (= 0.8 [s] ]) So acceptable (○)
(Judgment example 3) Since the average time of the active system communication time (= 0.8 [s]) and the standby system communication time (= 0.6 [s]) 0.7 [s] ≦ allowable time (= 0.8 [s]) Acceptable (○)
(Judgment example 4) Communication time of active system (= 0.8 [s]) + standby system communication time (= 0.6 [s]) = total time 1.4 [s]> allowable time (= 0.8 [s]), rejected (×)

FIG. 6 is a flowchart showing the processing contents of the placement destination determination unit 31. The placement destination determination unit 31 selects an appropriate DC base that places a service (VM) that the user wants to place based on each piece of information (user base information 11, placement service information 12, DC base information 21, communication facility information 22). To decide. Details will be described below.
As S101, when there are a plurality of users (UA, UB, UC...), The placement destination determination unit 31 orders the users in order to determine which user's cloud service is placed in order. This ordering process follows, for example, the following rules.
(Rule 1) Users with higher communication requirements for services registered in the arrangement service information 12 are selected first. The high required communication requirement means that more communication performance is required, for example, the allowable response time is short and the allowable communication time is short.
(Rule 2) When a tie is found in the above (Rule 1), a user having more user bases registered in the user base information 11 is selected first.
(Rule 3) In the case of a tie in the above (Rule 2), the user with the higher priority user bases registered in the user base information 11 is selected first.

For example, the users are ordered according to such rules as follows.
First-ranked user: Allowable communication time = 0.8 [s], out of 3 user bases, 2 places with high priority.
Second-ranked user: Allowable communication time = 0.8 [s], among the three user bases, one high priority.
Third-ranked user: Allowable communication time = 0.8 [s], and one of the two high-priority user locations.
Fourth-ranked user: Allowable communication time = 2.5 [s], among the three user bases, 0 places have high priority.
Then, according to the order determined in S101, the placement destination determination unit 31 sets the unselected and highest-ranked users as the current selection targets (S103) until there are no unselected users (until No in S102). Each process of S111-S123 is performed.

By determining the order of the users in accordance with the above (Rule 1) to (Rule 3), the users are arranged so as to be distributed at the respective positions as shown in FIG.
The square icons in the figure correspond to services that one icon uses for one user (has one or more user bases), and services with high required communication requirements in (Rule 1) are black squares (■). Services that do not have high required communication requirements are indicated by white squares (□).
The ellipse in the figure indicates a range within which the required communication requirement of the black square (■) can be satisfied. As shown in the figure, since the range of the ellipse is limited, by applying (Rule 1), a black square (■) can be placed in the ellipse first. Therefore, it is possible to suppress the black square (■) from falling outside the ellipse.

Returning to FIG. 6, the description will be continued. As S111, the placement destination determination unit 31 calculates the communication time from each user site to each DC site, as described with reference to FIG.
As S <b> 112, the placement destination determination unit 31 determines whether or not there is a DC base that is responsive to OK (that satisfies the necessary communication requirements) at all user bases for the same user. If Yes in S112, the process proceeds to S113, and if No, the process proceeds to S114. For example, in FIG. 5B, since the DC site D3 is responsive OK, the process proceeds to S113.
In S113, the placement destination determination unit 31 determines the DC base determined to be responsive in S112 as the placement destination.

  On the other hand, as S <b> 114, the placement destination determination unit 31 determines the DC base that is the placement destination based on the priority for each user base, as described in FIG. 5C. As described above, the placement destination determination unit 31 narrows down the user bases having a high priority set in advance, and sets the DC bases that can provide the user bases with better communication performance than other DC bases. Determine as placement destination. For example, the DC base that minimizes the sum of the communication times from the high-priority user bases may be determined, or the DC base that minimizes the difference between the communication times from the high-priority user bases may be determined. Good.

  As S 121, the placement destination determination unit 31 refers to the availability of the DC base information 21 for the placement destination DC base determined in S 113 or S 114, and needs the service to be placed described in the placement service information 12. Determine whether resources can be allocated. That is, it is determined whether or not there is a vacant performance on the physical server of the DC base of the arrangement destination. If Yes in S121, the process proceeds to S123, and if No, the process proceeds to S122.

  In S122, the placement destination determination unit 31 focuses on a user other than the user who is the current selection target in S103. The placement destination determination unit 31 changes the placed VM (and the cloud service operating on the VM) already used by the other user from the placement destination DC base determined in S113 or S114 to another DC base. By moving, the vacant performance is secured in the DC base of the movement source. That is, it is only necessary to perform a two-stage arrangement of a rearrangement process for an already placed VM and a placement process for a newly placed VM.

Note that the VM to be moved also needs to be responsive (satisfy required communication requirements) at the destination DC base. For this reason, the placement destination determination unit 31 may preferentially select a VM that is easy to move (a loose communication requirement such as a long allowable communication time) that is easy to move. The placement destination determination unit 31 also selects an appropriate destination DC base that satisfies the necessary communication requirements for the migration target VM, as in S112.
In addition, when the migration target VM cannot find a responsive DC base in S112, the placement destination determination unit 31 newly creates a migration target from a DC base that is another placement destination candidate based on S114. You may search VM.

FIG. 8 is a diagram in which a two-stage arrangement of services is performed on the cloud service construction network of FIG.
First, consider a case where the user to be selected this time uses the employee search DB (first row of the arrangement service information 12 in FIG. 3C) in S103. The necessary communication requirement of this employee search DB is “one-way communication time ≦ 0.8 seconds”, and as described with reference to FIG. 5B, only the DC site D3 is responsive OK (S112). However, it is assumed that the DC site D3 has no free performance (S121, No).

  Therefore, the placement destination determination unit 31 focuses on another cloud service “sales statistics DB” that is already placed on the DC site D3. The required communication requirement of this sales statistics DB is “uplink communication time ≦ 1.2 seconds and downlink communication time ≦ 2.4 seconds” in FIG. 3C, and any one-way required communication requirement of the employee search DB Loose than "one-way communication time ≤ 0.8 seconds". Therefore, the placement destination determination unit 31 moves the sales statistics DB to another DC site D2 that satisfies the necessary communication requirements of the sales statistics DB, thereby creating a space in the resource of the source DC site D3 (S122).

  And the arrangement | positioning instruction | indication part 32 arrange | positions employee search DB with respect to DC base D3 which is an arrangement | positioning destination of original employee search DB (S123). In the migration instruction in S122 and the placement instruction in S123, the placement instruction unit 32 notifies the virtual infrastructure system in the designated DC site D together with the placement service information 12, thereby providing a specific VM. You can direct the construction process.

  In the present embodiment described above, the location determination apparatus 1 provides communication quality that is necessary for providing services from any user base in consideration of communication performance from a plurality of distributed user bases to DC bases. A DC base at an appropriate position is selected as a cloud service placement destination. Therefore, the placement destination determination unit 31 of the placement destination determination device 1 reads the network topology from the communication facility information 22, estimates the communication performance from each user base to each DC base, as shown in FIG. As shown in FIG. 5B, the appropriate DC bases are narrowed down by evaluating each DC base based on the result of collation with the necessary communication requirements.

  In addition, since communication resources are limited, it may be difficult to determine an arrangement destination that satisfies all users due to an increase in the number of users. At that time, as shown in FIG. 5 (c), the DC base that preferentially satisfies the user base set with high priority is set as the placement destination, thereby greatly reducing the user satisfaction. Can be suppressed.

1 placement destination determination device (virtual machine placement destination determination device)
DESCRIPTION OF SYMBOLS 10 User information input part 11 User base information 12 Arrangement service information 20 DC information input part 21 DC base information 22 Communication equipment information 31 Arrangement destination determination part 32 Arrangement instruction part

Claims (4)

Position information on the network topology for each placement base that includes physical servers that are candidates for placement of virtual machines, and position information on the network topology for each user base that includes a user terminal that uses the virtual machine; A storage unit storing connection information of each communication device constituting the network topology;
By referring to the connection information of each communication device, the communication performance when communicating from each user base to each placement base is specified, and the communication performance satisfying the predetermined communication requirement from any user base the arrangement bases capable of providing, possess a placement destination determining section for determining a placement destination of the virtual machine,
The placement destination determination unit
When there is no placement base that can provide communication performance that satisfies the predetermined communication requirements from any of the user bases, communication performance that can be provided to the user base with a high priority set in advance is higher than other placement bases. Determine the location where the virtual machine is to be placed as the placement destination of the virtual machine,
When there are a plurality of users having one or more user bases, the order of users who arrange the virtual machines is determined according to the following rules:
(Rule 1) Users who require higher communication performance as the predetermined communication requirement are selected first.
(Rule 2) If there is a tie in (Rule 1) above, select the user with more user bases first
(Rule 3) A virtual machine placement destination determination device that selects a user with a higher number of user bases with a higher priority first when the score is the same in (Rule 2) . The placement destination determining unit refers to the connection information of each communication device, identifies the number of hops from the communication device housing each user base to the communication device housing each placement base, and this identified hop The virtual machine placement destination determination device according to claim 1, wherein a communication time proportional to the number is specified as a communication performance from each user base to each placement base. The placement destination determination unit, when the placement base determined as the placement destination of the predetermined virtual machine cannot provide a computer resource required by the predetermined virtual machine, another placement already determined on the determined placement base The virtual machine placement destination determination apparatus according to claim 1, wherein after moving a virtual machine to another placement base, the predetermined virtual machine is placed at the determined placement base. The virtual machine placement destination determination device includes a storage unit and a placement destination determination unit,
The storage unit includes location information on a network topology for each placement site including physical servers that are candidates for placement of virtual machines, and the network for each user site provided with a user terminal that uses the virtual machine. The location information on the topology and the connection information of each communication device constituting the network topology are stored respectively.
The placement destination determination section,
By referring to the connection information before Symbol each communication device, communication performance wherein identifying the communication performance when communicating to said each placement site from the user site, it satisfies a predetermined communication requirements from any user sites Is determined as the placement destination of the virtual machine ,
When there is no placement base that can provide communication performance that satisfies the predetermined communication requirements from any of the user bases, communication performance that can be provided to the user base with a high priority set in advance is higher than other placement bases. Determine the location where the virtual machine is to be placed as the placement destination of the virtual machine,
When there are a plurality of users having one or more user bases, the order of users who arrange the virtual machines is determined according to the following rules:
(Rule 1) Users who require higher communication performance as the predetermined communication requirement are selected first.
(Rule 2) If there is a tie in (Rule 1) above, select the user with more user bases first
(Rule 3) A virtual machine placement destination determination method in which a user having a higher number of user bases with a higher priority is selected earlier when the score is the same in (Rule 2) .

Images