JP6543090B2 - Load balancing device, load balancing method and program - Google Patents

Description

  The present invention relates to a load distribution device, a load distribution method, and a program.

2. Description of the Related Art Conventionally, a load distribution device that distributes requests from clients to a plurality of servers based on a set load distribution method has been widely used.
As a load distribution method, for example, a round robin method in which requests are distributed to each server in order is known.

  Further, as another load distribution method, Patent Document 1 describes a load distribution method in which weighting is performed according to the priority of packets, and packets are distributed in the order determined by the weighting.

WO 2010/101002

By the way, in the conventional round robin method and the load distribution method described in Patent Document 1, since requests are distributed to all servers, it is necessary to operate all the servers.
For this reason, even if it is not necessary to use the resources of all the servers to process the request from the client, the power consumption tends to be high because all the servers are used on average.
From such a thing, it is desired to save power of a plurality of servers connected to the load distribution device.

  Then, this invention is made in view of the situation mentioned above, Comprising: It aims at providing a suitable load distribution apparatus etc., when achieving power saving of a plurality of servers.

In order to achieve the above object, a load distribution device according to a first aspect of the present invention is:
A load balancer connected to a client and a plurality of servers via a network,
Measurement value acquisition means for acquiring a measurement value that is an indicator of the load on the plurality of servers;
Among the plurality of servers, a minimum number-of-servers acquiring unit for acquiring the minimum number of servers necessary for responding to the processing request per unit time from the client;
Based on the minimum number of servers obtained by the minimum server number obtaining means and the measured value obtained by the measured value obtaining means, a processing performance value capable of responding to a processing request per unit time by the plurality of servers Processing performance value acquisition means to be obtained;
Selection means for selecting a server that responds to the processing request from the client among the plurality of servers based on the processing performance value obtained by the processing performance value acquisition means and the fluctuation of the measurement value;
Distribution means for distributing the processing request from the client to the server selected by the selection means;
It is characterized by having.

Further, the measurement value acquisition unit may acquire the number of processing requests per unit time from the client as a measurement value that is an index of the load.
Further, the measurement value acquisition unit may acquire, as a measurement value serving as an index of the load, the number of processes processed in response to the processing request per unit time from the client.

Further, the minimum server number acquisition unit
Reference round trip time obtaining means for obtaining an average round trip time taken to receive response processing from the plurality of servers after transferring the processing request per unit time to the plurality of servers, as a reference round trip time;
An average round trip time acquiring unit for obtaining the average round trip time by reducing the number of the plurality of servers;
It is preferable to specify the number of servers in which the average round trip time exceeds a predetermined allowable round trip time calculated from the reference round trip time, and add the number of one server to the specified number of servers as the minimum number of servers.

Further, the minimum server number acquisition unit
The system comprises an average CPU utilization acquisition means for calculating an average CPU utilization from the CPU utilization of each server that processes the processing request per unit time in response.
It is preferable to specify the number of servers whose average CPU usage rate exceeds a predetermined upper limit rate and add the number of one server to the specified number of servers as the minimum number of servers.

  Further, the selection means responds to the processing request from the client among the plurality of servers so that there is a margin between the processing performance value obtained by the processing performance value acquisition means and the fluctuating measurement value. Choose a server to

  In addition, when the number of servers selected by the selection unit is two or more, the distribution unit may distribute the processing request from the client according to the performance ratio of the selected server.

  Further, the distribution unit is weighted to indicate a load distribution ratio so that the processing load includes a server lighter than other servers when two or more servers are selected so that the selection unit can make the margin. A coefficient may be assigned to each of the two or more servers, and the processing request from the client may be distributed based on the assigned weighting coefficient.

  Further, the distribution means dynamically changes the ratio of the weighting factor to be given to each of the servers according to the fluctuation of the measured value, and processing from the client based on the changed ratio of the weighting factor It is good to distribute the request.

  Further, it is preferable that the distribution unit equalizes the ratio of weighting factors to be given to the servers as the measurement value increases.

  Further, the processing performance value acquiring unit may recalculate the processing performance value when there is an increase or decrease in the number of the plurality of servers.

In order to achieve the above object, a load distribution device according to a second aspect of the present invention is:
Processing performance value acquisition means for acquiring processing performance values of a plurality of servers from the outside of the own device;
The processing request from the client among the plurality of servers is subjected to response processing based on the processing performance value acquired by the processing performance value acquiring means and the fluctuation of the measured value serving as an index of the load on the plurality of servers. Selection means for selecting a server;
Distribution means for distributing the processing request from the client to the server selected by the selection means;
It is characterized by having.

In order to achieve the above object, a load distribution method according to a third aspect of the present invention is:
A measurement value acquisition step of acquiring a measurement value that is an index of the load on the plurality of servers;
Among the plurality of servers, a minimum number of servers obtaining step for obtaining a minimum number of servers necessary for responding to processing requests per unit time from clients.
Based on the minimum number of servers obtained in the minimum server number obtaining step and the measured value obtained in the measured value obtaining step, a processing performance value capable of responding to a processing request per unit time by the plurality of servers Processing performance value acquisition step to be determined;
A selection step of selecting a server to which the processing request from the client is to be subjected to response processing among the plurality of servers based on the processing performance value obtained in the processing performance value acquisition step and the fluctuation of the measurement value;
A distribution step of distributing the processing request from the client to the server selected in the selection step;
It is characterized by having.

In order to achieve the above object, a program according to a fourth aspect of the present invention is
A computer connected via a network to a client and multiple servers,
Measurement value acquisition means for acquiring measurement values that are indicators of the load on the plurality of servers,
Among the plurality of servers, a minimum number-of-servers acquiring unit for acquiring the minimum number of servers necessary for responding to the processing request per unit time from the client,
Based on the minimum number of servers obtained by the minimum server number obtaining means and the measured value obtained by the measured value obtaining means, a processing performance value capable of responding to a processing request per unit time by the plurality of servers Processing performance value acquisition means to
Selecting means for selecting a server to which the processing request from the client is to be subjected to response processing among the plurality of servers based on the processing performance value obtained by the processing performance value acquiring means and the fluctuation of the measured value;
Distribution means for distributing the processing request from the client to the server selected by the selection means;
Is a program to function as

  According to the present invention, power saving of a plurality of servers can be achieved.

It is a figure showing roughly the composition of a 1st embodiment of this invention. It is a block diagram which shows the hardware constitutions of the load distribution apparatus shown in FIG. It is a functional block diagram of the load distribution apparatus shown in FIG. It is a flowchart which shows the flow of the minimum server number acquisition process which concerns on 1st Embodiment. It is a figure which shows an example of the average round trip time with respect to each server number. It is a functional block diagram of a load distribution device concerning a 2nd embodiment. It is a flowchart which shows the flow of the minimum server number acquisition process which concerns on 2nd Embodiment. It is a figure which shows an example of the average CPU utilization with respect to each server number. It is a figure which shows the amount of transactions, the number of operation servers, and the example of provision of the weighting coefficient provided to each server. (A) is a graph which shows the example of the relationship between the load factor of a server, and the power consumption of a server. (B) is a figure for demonstrating the relationship between a weighting coefficient and power consumption.

First Embodiment
Hereinafter, a first embodiment of the present invention will be described based on the drawings.
As shown in FIG. 1, the load distribution apparatus 100 is connected to a plurality of clients 200 and a plurality of servers (1 to 5) via a network. The network between the load distribution apparatus 100 and the plurality of servers is, for example, a LAN (Local Area Network), and the network between the load distribution apparatus 100 and the plurality of clients 200 is, for example, the Internet or LAN.
In the following description, in the case where it is not necessary to specifically identify and explain each server, it will be appropriately referred to as a plurality of servers, all servers, each server and the like. Further, in this embodiment, in order to facilitate understanding, it is assumed that the number of servers is five and the performance and power consumption of each server are the same.

  The load distribution apparatus 100 is a load balancer that relays processing requests from a plurality of clients 200 transmitted via the network and distributes the processing requests to each server. In this embodiment, the load distribution apparatus 100 distributes the processing request to the selected server without distributing the processing request to all the servers in order to reduce the power consumption. This point will be described later.

The plurality of servers are web servers that provide data such as HTML (HyperText Markup Language) documents and images as response processing in response to processing requests from a plurality of clients 200.
Each of the plurality of clients 200 is a PC (Personal Computer) that receives a user's input operation and transmits a processing request. Also, each of the plurality of clients 200 displays an HTML document, an image, etc. on display means such as a display based on data provided from each server. Here, the processing request is, for example, an http (HyperText Transfer Protocol) request, and the response processing is, for example, an http response.
Although only a plurality of clients 200 (clients 210, 220 and 230) are illustrated for the sake of convenience, several hundreds and several thousands are actually connected to the load balancer 100 via the Internet, and a plurality of processing requests are each server Distributed to

Next, the hardware configuration of the load distribution apparatus 100 will be described with reference to FIG. As shown in FIG. 2, the load distribution apparatus 100 includes a central processing unit (CPU) 101, a flash read only memory (ROM) 102, a random access memory (RAM) 103, and a network I / F (interface) 104. .
Among them, the CPU 101 is a central processing unit that controls the entire load distribution apparatus 100.
The flash ROM 102 is a non-volatile memory storing a program related to load distribution executed by the CPU 101.

A random access memory (RAM) 103 is a volatile memory used as a work area when the program executed by the CPU 101 is temporarily expanded and the CPU 101 performs processing related to load distribution.
A network I / F 104 is an interface for the load distribution apparatus 100 to communicate with a plurality of servers and a plurality of clients 200 via a network.

Here, the CPU 101 reads out a program related to load distribution in the flash ROM 102, develops it in the RAM 103, and controls the load distribution apparatus 100 according to the program to realize the functions of the respective parts as shown in FIG. Can.
As functions, as shown in FIG. 3, the number of transactions acquiring unit 10, the minimum number of servers acquiring unit 20, the maximum number of transactions acquiring unit 30, the server selecting unit 40, and the distributing unit 50 are provided.

First, the transaction number acquisition unit 10 measures the number of transactions (TPS: Transaction Per Sec) that has been processed in response to processing requests per second from a plurality of clients 200, and acquires the number of transactions. Here, a transaction refers to a series of processing in which processing up to response processing request is integrated as an integral part.
For example, the transaction number acquisition unit 10 transfers one processing request received from the client 210 to the server 1 and sets it as one transaction when receiving the response processing from the server 1. In this manner, the transaction number acquisition unit 10 transfers the processing requests from the plurality of clients 200 to each server, and counts the transactions when the response processing from each of the servers is received.

  Here, the number of transactions can be said to be a measurement value that is an indicator of the load on a plurality of servers. Therefore, the higher the number of transactions, the higher the load, and the lower the load, the lower the load. In this embodiment, the number of transactions (TPS) will be described as an example of a measurement value as an index of load. The number of transactions is also called the number of processes.

In this embodiment, before selecting the load distribution destination (that is, when all the servers are operating), the transaction number acquisition unit 10 responds to the processing request per second from the plurality of clients 200, and all the servers Measure and acquire the number of transactions processed in response. On the other hand, after selecting the load distribution destination (that is, when a dormant server is included), the number-of-transactions acquiring unit 10 responds to the processing request per second from the plurality of clients 200, the number of transactions that the active server responded Measure and get.
The number of transactions is in principle the same as the number of processing requests, with the exception that the response processing could not be performed due to a failure or the like. The transaction number acquisition unit 10 corresponds to a measurement value acquisition unit.

Next, the minimum number-of-servers acquisition unit 20 obtains the minimum number of servers necessary for responding to the processing request per second from the plurality of clients 200 among the plurality of servers.
Specifically, first, the reference round trip time acquiring unit 21 transfers the processing request per second to a plurality of servers, and then the average round trip time taken to receive the response processing from the plurality of servers is a reference round trip Ask for time.
Next, the average round trip time acquiring unit 22 obtains the average round trip time by reducing the number of servers, and compares the calculated average round trip time with a predetermined allowable round trip time calculated from the reference round trip time. .

Next, the minimum number-of-servers acquisition unit 20 specifies the number of servers whose average round-trip time exceeds the predetermined allowable round-trip time, and sets the number of one server as the minimum number of servers. Specific processing for obtaining the minimum number of servers by the minimum server number acquisition unit 20 will be described later.
The minimum server number acquisition unit 20 corresponds to the minimum server number acquisition unit, the reference round trip time acquisition unit 21 corresponds to the reference round trip time acquisition unit, and the average round trip time acquisition unit 22 corresponds to the average round trip time acquisition unit.

Next, the maximum number-of-transactions acquisition unit 30 sets the number of servers per second based on the minimum number of servers obtained by the minimum-server number acquisition unit 20 and the number of transactions which is the measured value acquired by the transaction number acquisition unit 10. Determine the maximum number of transactions (maximum TPS) that can process the processing request in response to each request.
Specifically, the maximum number of transactions acquiring unit 30 determines the maximum number of transactions for all five servers based on the ratio between the determined minimum number of servers and the number of transactions processed by all the servers measured before selecting the load distribution destination. Find the number. A specific example of determining the maximum number of transactions will be described later.
The maximum number of transactions indicates the maximum processing performance value of the plurality of servers as a whole. Also, the maximum transaction number acquisition unit 30 corresponds to processing performance value acquisition means.

Next, the server selection unit 40 responds to the processing request from the plurality of clients 200 among the plurality of servers based on the maximum number of transactions which is the maximum processing performance value thus obtained and the fluctuation of the number of transactions measured. Select a server to process.
In other words, the server selection unit 40 selects one of the plurality of servers according to the ratio of the transaction number (TPS) acquired by the transaction number acquisition unit 10 to the maximum transaction number (maximum TPS) calculated by the maximum transaction number acquisition unit 30. A server that responds to the processing requests from the plurality of clients 200 is selected. That is, the server selection unit 40 determines the number of servers to be distributed.
Specifically, the server selection unit 40 dynamically increases or decreases the number of servers to be subjected to response processing according to the fluctuation of the measured number of transactions. For example, when the ratio of TPS to the maximum TPS is small, that is, when the measured number of transactions is small, it is not necessary to operate all the servers, so the server selection unit 40 selects two servers, server 1 and server 2, for example. . In the selection, an arbitrary server is selected among a plurality of servers. Further, the server selection unit 40 corresponds to selection means.

  Next, the distribution unit 50 distributes the processing requests from the plurality of clients 200 to the server selected by the server selection unit 40. Although the method of distribution is arbitrary, for example, it is equally distributed according to the round robin method. The distribution unit 50 corresponds to distribution means.

As described above, in the load distribution apparatus 100 described with reference to FIGS. 1 to 3, for example, one characteristic point is to obtain the minimum number of servers, and based on the obtained minimum number of servers, the maximum processing performance of all servers It is the point which asks for a value. Therefore, the minimum server number acquisition processing related to this point will be described below with reference to the flowchart of FIG.
This process is performed immediately after activating the load distribution apparatus 100 or every predetermined time (for example, once a week) after starting operation of a plurality of servers. Further, it is assumed that the number of transactions measured and acquired by the transaction number acquisition unit 10 is 1000 TPS.

  First, the minimum number-of-servers acquisition unit 20 sets the number of servers as the number of servers (step S11). In this embodiment, the minimum server number acquisition unit 20 sets the number of servers as five.

  Next, the reference round trip time acquisition unit 21 measures a reference round trip time (step S12). Specifically, the reference round trip time acquiring unit 21 transfers the processing request per second to a plurality of servers, and then takes the average round trip time taken to receive the response processing from the plurality of servers as the reference round trip time. Do. In the case where the average round trip time is, for example, 1 millisecond, as shown in FIG. 5, the average round trip time (reference round trip time) for five vehicles is 1 millisecond.

  Next, the minimum number-of-servers acquisition unit 20 calculates the allowable round trip time (step S13). Specifically, the minimum server number acquisition unit 20 multiplies the reference round trip time by a predetermined coefficient to calculate an allowable round trip time. For example, 1 ms, which is a reference round trip time, is multiplied by a coefficient 2 to obtain an allowable round trip time of 2 ms. The coefficient can be set by the management user of the load distribution apparatus 100 to a desired value.

  Next, the average round trip time acquisition unit 22 sets the number of servers as the number of servers minus 1 (step S14). In this embodiment, the minimum server number acquisition unit 20 sets the number of servers to four.

Next, the average round trip time acquisition unit 22 determines whether the number of servers = 0 (step S15). Here, since the number of servers is four and not zero (step S15; No), the average round trip time acquiring unit 22 measures the average round trip time (step S16).
Specifically, the average round trip time acquiring unit 22 obtains an average round trip time of 1.1 msec when the number of servers is four (see FIG. 5).
Here, in general, when the number of servers decreases, the number of server servers responding to processing requests increases, so the average round trip time becomes longer.

  Next, the minimum server number acquisition unit 20 determines whether the average round trip time is equal to or less than the allowable round trip time (step S17). Here, since the average round trip time of 1.1 seconds is equal to or less than the allowable round trip time of 2 ms (step S17; Yes), the process returns to step S14, reduces the number of servers by one, and performs steps S15 to S17 again. As described above, in the processes of steps S14 to S17, the number of servers is reduced one by one, and it is determined each time whether the average round trip time is equal to or less than the allowable round trip time. Then, the processes in steps S14 to S17 are repeated until the average round trip time exceeds the allowable round trip time (step S17; No).

In the example of FIG. 5, when the number of servers is one, the average round trip time of 2.5 msec exceeds the allowable round trip time of 2 msec (step S17; No). In this case, the minimum server number acquisition unit 20 sets the minimum server number = server number + 1 (step S18), and ends the process. In the case of the example of FIG. 5, the minimum number-of-servers acquiring unit 20 sets the minimum number of servers to two immediately before the number of servers exceeding the allowable round trip time, and ends the processing.
In this case, since the minimum server number is two and the number of transactions (processing number) measured is 1000 TPS, the maximum TPS is 1000 TPS × 5 units / 2 units = 2500 TPS. Therefore, the number of transactions that can be processed per server is 500 TPS.

On the other hand, unlike the example of FIG. 5, even when the number of servers is one, if the average round trip time is equal to or less than the allowable round trip time (step S17; Yes), the process returns to step S14 and the number of servers is zero. (Step S15; Yes), the minimum number-of-servers acquiring unit 20 sets the minimum number of servers to one (Step S19), and ends the process.
In this case, since the minimum number of servers is one and the number of transactions (number of processing) measured is 1000 TPS, the maximum TPS is 1000 TPS × 5/1 = 5000 TPS. Therefore, the number of transactions that can be processed per server is 1000 TPS.

  In this way, once the maximum number of transactions (maximum TPS) has been obtained, the server selection unit 40 determines the number of servers required according to the fluctuation of the number of transactions (TPS), and the number of servers corresponding to that number is determined. select. For example, if the maximum TPS is 2500 TPS and the TPS is 1000 TPS, two servers (for example, servers 1 and 2), and if the measured TPS is 1100, three servers (for example, servers 1 to 3) are selected. It will be.

As described above, in the process of FIG. 4, the load distribution apparatus 100 includes the functions of the transaction number acquisition unit 10, the minimum server number acquisition unit 20, and the maximum transaction number acquisition unit 30, so that one server is included from the plurality of servers. The average round trip time is measured, the average round trip time is measured, the number of servers whose specified round trip time exceeds the allowable number of servers, and the number of servers specified is added to the number of one server (immediately The maximum number of transactions can be determined with the number of servers as the minimum number of servers.
Therefore, it is possible to obtain the maximum processing performance value required to determine the number of servers used. Therefore, there is no need for the user to obtain the maximum processing performance value. In addition, servers that have not been selected can be put into the idle state. For this reason, power saving of a plurality of servers connected to the load distribution apparatus 100 can be achieved while saving the trouble of the user.

  The average round trip time acquiring unit 22 may obtain the average round trip time by reducing the number of servers, and may obtain the delay time that is the difference between the calculated average round trip time and the reference round trip time. In this case, the minimum number-of-servers acquisition unit 20 specifies the number of servers whose delay time exceeds the predetermined allowable delay time, and sets the number of one server as the minimum number of servers to which the specified number of servers is added.

Specifically, first, instead of calculating the allowable reciprocation time in step S13, the minimum number-of-servers acquiring unit 20 calculates the allowable delay time from the reference reciprocation time (for example, when the reference reciprocation time is 1 ms, 1 m Calculate the allowable delay time on the basis of seconds as 1 millisecond). Next, the delay time (difference between the average round trip time and the reference round trip time) when the number of servers is reduced (four, three, two, etc.) is sequentially determined, and the delay time and the allowable delay time determined each time The number of servers whose delay time exceeds the allowable delay time (for example, if the delay time of one is 1.5 ms, which exceeds the allowable delay of 1 Assume the number of servers.
As a result, power consumption of a plurality of servers connected to the load distribution apparatus 100 can be reduced by using the allowable delay time instead of the allowable reciprocation time to save the same effect as that of the first embodiment, that is, saving the trouble of the user. Can be

Second Embodiment
Next, a second embodiment of the present invention will be described with reference to FIGS.
The second embodiment is different from the first embodiment in that the load distribution apparatus 100 ′ obtains the minimum number of servers based on the average CPU usage rate. Therefore, the following description will be focused on this difference.
In the second embodiment, the hardware configuration of the load distribution apparatus 100 'is the same as that of the load distribution apparatus 100 of the first embodiment. Further, as to the function of the load distribution apparatus 100 ', as shown in FIG. 6, an average CPU usage rate acquiring unit 23 is provided instead of the reference reciprocating time acquiring unit 21 and the average reciprocating time acquiring unit 22 shown in FIG. Is different.

  The average CPU usage rate acquisition unit 23 obtains an average CPU usage rate from the CPU usage rates of the servers that process response requests per second. For example, when the measured number of transactions is 1000 TPS, the average CPU usage rate acquisition unit 23 obtains the average CPU usage rate with different numbers of servers (five, four, three, etc.).

Here, the method of acquiring the CPU usage rate from each server is arbitrary. For example, a server monitoring module called SNMP (Simple Network Management Protocol) agent is installed in advance in each server, and the load balancer 100 'issues an SNMP get command to the SNMP agent to use the CPU. You can get the rate.
Note that the CPU usage rate may be acquired by another method. Further, the average CPU usage rate acquisition unit 23 corresponds to an average CPU usage rate acquisition unit.

  Next, the flow of the minimum server number acquisition process according to the second embodiment will be described with reference to FIG. The implementation timing is the same as in the first embodiment. Further, it is assumed that the number of transactions measured and acquired by the number-of-transactions acquisition unit 10 is 1000 TPS.

  First, the minimum number-of-servers acquisition unit 20 sets the number of servers as the number of servers (five servers) (step S21). Next, the average CPU usage rate acquisition unit 23 calculates an average CPU usage rate (step S22). For example, the average CPU usage rate acquisition unit 23 calculates an average CPU usage rate of five servers as 40% (see FIG. 8).

  Next, the minimum number-of-servers acquisition unit 20 determines whether the average CPU usage rate is less than or equal to the upper limit rate (step S23). The upper limit rate is set in advance by the management user of the load distribution apparatus 100 '. In the second embodiment, the upper limit rate is 80% as an example.

  Here, if the average CPU usage rate of the five servers is 85% different from that of FIG. 8, the average CPU usage rate exceeds the upper limit rate (step S23; No), the minimum server number acquisition unit 20 is the minimum server number = The number of servers (five) is set (step S30), and the process is ended. If the upper limit rate is exceeded for all servers, it is necessary to cope with all servers. In this case, the maximum TPS of all five servers is 1000 TPS (the number of processes that can be processed per server is 200 TPS).

  On the other hand, if the average CPU usage rate of five servers is 40% in FIG. 8, the average CPU usage rate is below the upper limit rate (step S23; Yes), the number of servers is reduced by one (step S24). Do. Next, since the number of servers (4) is not 0 (step S25; No), the average CPU usage rate acquiring unit 23 calculates an average CPU usage rate of four servers (step S26). Next, the minimum number-of-servers acquisition unit 20 determines whether the average CPU usage rate of four servers is less than or equal to the upper limit rate (step S27).

  Here, in the case of the example of FIG. 8, since the average CPU utilization rate of four units is 50%, the upper limit rate of 50% is 50% or less (step S27; Yes), and the process returns to step S24. Steps S24 to S27 correspond to steps S14 to S17 in the first embodiment, and each time the number of servers is reduced by one until the average CPU usage rate exceeds the upper limit rate (step S27; No). The average CPU usage rate is calculated, and it is determined whether the calculated average CPU usage rate is less than or equal to the upper limit rate.

In the case of the example of FIG. 8, the CPU usage rate for one server is 85% and exceeds the upper limit rate 80% (step S27; No), the minimum number of servers acquiring unit 20 is the minimum number of servers = The number of servers + 1 is set (step S28), that is, the minimum number of servers is set to two, and the process is ended.
In this case, since the TPS measured when the minimum number of servers is two is 1000 TPS, the maximum TPS is the same 2500 TPS as in the first embodiment.

On the other hand, unlike the example of FIG. 8, when the CPU usage rate for one server is 75%, the upper limit rate does not exceed 80% (step S27; Yes), the process returns to step S24 and the number of servers = 0 (Step S25; Yes), the minimum server number acquisition unit 20 sets the minimum number of servers to one (step S29), and ends the process.
In this case, since the TPS measured with one minimum server is 1000 TPS, the maximum TPS is the same 5000 TPS as the first embodiment.

As described above, in the second embodiment described with reference to FIGS. 6 to 8, the load distribution apparatus 100 ′ includes the function of the average CPU usage rate acquisition unit 23, thereby reducing the minimum number of servers based on the allowable round trip time. Instead of the method to be determined, the minimum number of servers is determined based on whether the average CPU utilization exceeds the upper limit rate. Specifically, the load balancer 100 'identifies the number of servers whose average CPU usage rate exceeds a predetermined upper limit rate, and adds the number of one server to the identified number of servers as the minimum number of servers. The number of transactions can be determined.
For this reason, in addition to the ability to automatically obtain the maximum processing performance value as in the first embodiment, the average CPU usage rate is used, and therefore the minimum number of servers can be obtained as compared to the case where the allowable round trip time is used. It can be determined that the processing performance has reached the limit with higher accuracy.

  This is the end of the description of each embodiment, but it goes without saying that the specific configuration of the load distribution apparatus 100 (100 ') and the contents of processing are not limited to those described in the above embodiments.

For example, in the first embodiment, the minimum number of servers is obtained based on the round trip time of the processing request such as the http request, but the present invention is not limited to this. For example, the minimum number of servers may be determined using Ping used in network communication confirmation instead of the processing request.
Specifically, the average round trip time from when the load distribution apparatus 100 issues an IP (Internet Protocol) packet and transmits it to each server until it receives a response is used. At this time, as in the first embodiment, based on the average round trip time for five servers, the number of servers is reduced by one at a time to four, three, two, and so on to obtain the average round trip time. Ask for As a result, the minimum server count can be obtained by diverting the alive monitoring of servers using Ping by the load distribution apparatus 100.

Moreover, in the minimum server number acquisition process of the first embodiment, although it has been described on the premise that each server responds to the processing request and the round trip time can be determined, it is determined that the round trip time is not received. Sometimes you can not.
In such a case, a time-out time may be set in advance in the load distribution apparatus 100, and if there is no response within that time, the minimum server number acquisition process may be forcibly terminated. Alternatively, the minimum number of servers may be set as the minimum number of servers, for example, the number of servers immediately before time-out did not occur (for example, three if there were responses with three but no responses with two).

  Further, in the first and second embodiments, the minimum number of servers is obtained using a predetermined threshold (the allowable round trip time or the upper limit rate of the average CPU usage rate), but the present invention is not limited to this. For example, when each server can not process the response process to the process request within a predetermined time, and when each server measures the CPU usage rate by itself and exceeds a predetermined upper limit rate, an alert such as an error message can be load-balanced. It may be sent to 100 (100 '). As a result, the load distribution apparatus 100 (100 ') can calculate the minimum number of servers based on the received error message instead of determining the minimum number of servers depending on whether the predetermined threshold is exceeded or not. The load can be reduced.

Further, in the first and second embodiments, even if only one server does not exceed the allowable round trip time or the upper limit rate of the CPU utilization, one server is set as the minimum number of servers (see steps S19 and S29). It is not limited to. For example, it is also assumed that the number of transactions (TPS) when the minimum server number acquisition process is started is very small, that is, the number of processing requests from a plurality of clients 200 is very small.
In such a case, assuming that the minimum number of servers is one, it may not be possible to cope with the case where the number of transactions increases later, so instead of setting the minimum number of servers to one (Steps S19 and S29) Alternatively, the minimum server number acquisition process may be restarted when the number of transactions has increased. For example, it is conceivable to restart when the number of transactions reaches a predetermined number.

In the first and second embodiments, the minimum number of servers acquisition process is performed every predetermined time after starting the load balancer 100 (100 ′) or after starting operation of a plurality of servers (for example, in one week) Although it was decided to be implemented once, etc., it is not limited to this. For example, the minimum server number acquisition process may be performed at the timing when the number of servers increases or decreases. As a result, it is possible to obtain the maximum processing performance value again at the timing when the number of servers has changed, such as system migration.
Also, not only when the number of servers has increased or decreased, but also when there is a change in server performance due to server replacement, or when the server content has changed, etc., the minimum number of servers is acquired similarly. It is also good.

In the first and second embodiments, the power consumption of each server is described to be the same, but the present invention is not limited to this. For example, the power consumption of each server may be different. In this case, when the load distribution apparatus 100 (100 ') stores the power consumption of the servers 1 to 5 in advance and selects a load distribution destination server, it is preferable to select a server with a small power consumption first. Good. As a result, the hibernation server inevitably has high power consumption, and power saving can be enhanced.
Also, a server with a large performance to power ratio may be selected preferentially. The performance to power ratio indicates, for example, a value obtained by dividing the performance of the server by the power consumption. In this case, the performance to power ratios of the servers 1 to 5 are stored in advance, and when selecting a load distribution destination server, it is preferable to preferentially select a server having a small performance to power ratio.

  In the first and second embodiments, the description has been made on the premise that the performance of each server is the same, but the present invention is not limited to this. For example, the performance of each server may be different. In this case, when there are two or more servers selected by the server selection unit 40 of the load distribution apparatus 100, processing requests from a plurality of clients 200 may be distributed according to the performance ratio of the selected server. For example, when the selected server is the servers 1 and 2, the processing request may be distributed to the servers 1 and 2 according to the weighting ratio according to the performance. Thereby, the response processing can be performed in consideration of the performance of the server.

  Further, in the first and second embodiments, the number of transactions (TPS) has been described as an example of the measurement value serving as the load index, but the present invention is not limited to this. For example, the number of processing requests per unit time from a plurality of clients 200 may be used as a measurement value serving as a load indicator. In this case, the load distribution apparatus 100 (100 ′) acquires, for example, the number of http requests per second as a measurement value. Then, the load distribution apparatus 100 (100 ') is able to process the maximum number of requests that a plurality of servers can respond to http requests per second based on the obtained minimum number of servers and the obtained number of http requests. Calculate as a value. Then, the server to be operated may be selected based on the calculated maximum number of requests and the fluctuation of the measured number of requests.

Further, in the first and second embodiments, although the case where the load distribution apparatus 100 (100 ') obtains the maximum processing performance value of the plurality of servers as a whole has been described as an example, the present invention is not limited thereto. For example, the load distribution apparatus 100 (100 ′) may acquire the maximum processing performance value (for example, the maximum number of transactions, etc.) from the outside of the own apparatus.
Specifically, the minimum number of servers is obtained outside the own apparatus, and the processing performance value is obtained in advance from the calculated minimum number of servers and the measured value (for example, the number of transactions) measured. Then, the load distribution apparatus 100 (100 ') receives and stores processing performance values obtained outside the own apparatus in advance via a network or the like. Then, the load distribution apparatus 100 (100 ′) dynamically selects a load distribution destination server based on the processing performance value acquired from the outside and the fluctuation of the measured value (for example, the number of transactions).
As a result, the load distribution apparatus 100 (100 ') need not have the function (the number of transactions acquisition unit 10, the minimum number of servers acquisition unit 20, the maximum number of transactions acquisition unit 30) for obtaining the processing performance value by itself. Therefore, power saving of a plurality of servers can be achieved while suppressing the processing load.

Third Embodiment
In the description of the first embodiment, it has been described that the server selection unit 40 decides to distribute transactions to some of a plurality of servers on the assumption that there is no rapid increase in transactions (processing requests). However, when the distributed state is close to the processing performance value of each server (when there is no processing margin), if there is a sharp increase in transactions, a transaction in which the response processing can not be performed occurs. This is because it takes a predetermined processing time to add a distribution destination server, so it is impossible to process transactions exceeding the processing performance value acquired within the processing time. In the third embodiment, a technique will be described in which the server selection unit 40 determines the number of servers of distribution destinations in consideration of a rapid increase in transactions.

  Further, in the description of the first embodiment, the distribution unit 50 equally distributes the acquired transactions according to the round robin method to the server selected by the server selection unit 40. In general, in order to reduce the amount of unprocessed transactions which can not be processed in response to a rapid increase in transactions, it is possible to reduce the number of unprocessed transactions by equally distributing the transactions to each server. On the other hand, when allocating the same amount of transactions to a plurality of servers with the same performance, it is better to assign weighting factors to servers rather than allocating transactions evenly, that is, transactions can be performed so that a server with a light processing load can be made. If distributed, overall power consumption can be reduced. Even when operating with a margin in server processing power, it is preferable to reduce the overall power consumption by considering the transaction distribution method. In the third embodiment, a distribution technique to which this weighting coefficient is assigned will also be described.

  In the present embodiment, in order to facilitate understanding, the number of servers in the entire system shown in FIG. 1 will be described as three servers, server 1 to server 3. The processing performance value of each server is 1000 TPS.

  In the third embodiment, the hardware configuration of the load distribution apparatus 100 is the same as that of the load distribution apparatus 100 of the first embodiment. The functional configuration of the load distribution apparatus 100 is also the same as that shown in FIG. However, the server selection unit 40 and the distribution unit 50 have the functions described below.

  First, the function in which the server selection unit 40 selects the number of servers having a processing margin for a rapid increase in transaction amount will be described. The server selection unit 40 is between the processing performance value acquired by the maximum transaction number acquisition unit 30 (corresponding to the processability value acquisition means) and the transaction amount (corresponding to the measurement value) acquired by the transaction number acquisition unit 10. Select a server that responds to transactions (request processing) from clients so that there is a processing margin. Specifically, when the transaction amount acquired by the transaction number acquisition unit 10 reaches a predetermined ratio of the maximum transaction amount obtained by the maximum transaction number acquisition unit 30, one server number to which the transaction is distributed is added.

  In the present embodiment, when a transaction exceeding 50% of the processing performance value of the server is acquired, the server selection unit 40 adds one selected server number. In the example shown in FIG. 9, when two servers, server 1 and server 2, are operating, the processing performance value of one server is 1000 TPS, so the server selection unit 40 has 50% of the transaction amount of 2000 TPS. If 1000TPS is exceeded, the number of servers for distribution will be increased to three.

  Next, a function will be described in which the distribution unit 50 assigns a weighting coefficient to the server to which the transaction is distributed, and distributes the transaction based on the weighting coefficient. The weighting coefficient assignment method is applied to other servers so that a server with a light processing load can do it. In this embodiment, the weighting factor is dynamically changed and assigned in five stages according to the transaction amount (corresponding to the measurement value). Specifically, the ratio of the weighting factor to be assigned is increased as the amount of transaction decreases, and the weighting factor is assigned so that a server with a light processing load can be provided. Then, as the transaction amount increases, the weighting factors are dynamically changed so as to become equal gradually. This is to harmonize the increase of the margin of each server against the rapid increase of transactions and the reduction of the power consumption of the whole server described later.

  In the example illustrated in FIG. 9, when the transaction amount is less than 10% of the processing performance value of one server obtained by the maximum number of transactions acquiring unit 30, the distribution unit 50 sends 10% of the transaction amount to the server 2. The distribution is performed, and the remaining transactions are distributed to the server 1. That is, the distribution ratio of the transaction amount to the server 1 and the server 2 is (9: 1). When the transaction amount is 10% or more and less than 20% of the maximum number of transactions, the distribution unit 50 distributes 20% of the transaction amount to the server 2 and distributes the remaining transactions to the server 1. That is, the distribution ratio of the transaction amount to the server 1 and the server 2 is (8: 2). Similarly, the distribution unit 50 dynamically changes the distribution ratio of the transaction amount to the server 1 and the server 2 to (7: 3) (6: 4) (5: 5) according to the transaction amount. Let me set it. In this way, as the transaction amount (corresponding to the measured value) increases, the distribution ratios (ratios of weighting factors) given to the respective servers are made to be gradually equal.

  When the transaction amount exceeds 1000 TPS, the server selection unit 40 selects three servers, server 1, server 2 and server 3, as distribution destination servers. In this case, the processing amount of the two servers that are operating first (500 TPS × 2 units = 1000 TPS) are subtracted from the acquired transaction amount. Then, if the remaining transaction amount corresponds to 0 to 10% of the processing performance value of the server 3, 10% of the transaction amount is distributed to the third server 3, and the remaining transactions are equally distributed to the server 1 and the server 2. Distribute. If the remaining transaction amount is 10% or more and less than 20% of the processing performance value of server 3, 20% of the transaction amount is distributed to the third server 3, and the remaining transactions are equally distributed to server 1 and server 2. . Likewise, weighting coefficients are assigned as shown in FIG. 9 according to the transaction amount.

The weighting coefficient (X: X: S) assignment method according to the present embodiment is as follows. The parameter S is a parameter determined by the transaction amount for the processing performance value of one server, and is a natural number of 1 to 5. As described above, in the case of less than 10%, S = 1, and in the case of 10% or more and 20% or less, S = 2.
X = (5N-S) / (N-1) .. Formula (1)
S: 5-step parameter
N: Number of servers selected by the server selection unit 40 For example, in the case of N = 2, the distribution ratio of the server 1 and the server 2 is (9: 1) (8: 2) (7: 3) (7: 3) depending on the transaction amount. It changes to 6: 4) (5: 5).

  The reason for assigning weighting factors to servers and distributing transactions according to the weighting factors is to reduce the power consumption of the entire server. Since the server increases the clock frequency of the CPU and improves the processing capacity as the server load factor is higher, the power consumption often increases as the server load factor is higher. Therefore, as for the relationship between the load factor of the server and the power consumption, as the load factor of the server is lighter, the power consumption tends to decrease as shown by the graph shown in FIG. In addition, when the load factor (for example, about 50%) is reached, the area operates with best effort, so the rate of increase in power consumption tends to be moderate when the processing load becomes heavier. FIG. 10B is an example showing the difference in power consumption of the entire server according to the distribution ratio when the same amount of transactions is distributed to two servers by the distribution ratio of (X: S). The column of “power consumption index” in FIG. 10B is an example of a relative index of power consumption for each server load factor when the power consumption of the server when the server load factor is 50% is 100. As shown in the column "sum of power consumption index" in FIG. 10 (b), when the same amount of transactions is distributed to two servers, the transactions are processed so that a server with a heavy processing load and a server with a light processing load can be created. If the distribution ratio is biased, the power consumption of the entire server decreases. Even when the same amount of transactions is distributed to three or more servers, the power consumption of the entire server can be reduced by reducing the processing load of the additionally operated server, as compared to the case where the transactions are evenly distributed.

  For example, when operating at 300 TPS, if the transaction volume sharply increases to 1200 TPS, and if only one server 1 is operating, a 200 TPS transaction will be outstanding. However, as shown in FIG. 9, when the transactions are distributed to server 1 and server 2 (8: 2) and operated as two machines, transactions of 960 TPS and 240 TPS are distributed to server 1 and server 2, respectively, which are not processed yet. Transactions do not occur. Furthermore, as described above, the power consumption of the entire server can be reduced as compared to the case of equally distributing transactions.

The method of providing the margin described in the third embodiment is not limited to the method described above. For example, servers may be added with a threshold of 60% or 70% of the processing performance value of the server. As the processing performance value of the server, the average round trip time described in the first embodiment may be used, or the average CPU usage rate described in the second embodiment may be used. Further, the server selection unit 40 may select the number of distribution destination servers so as to secure a predetermined transaction amount as a margin.

Further, the method of applying weighting coefficients described in the third embodiment is not limited to the example described in the third embodiment. A weighting factor may be assigned to determine the distribution ratio of transactions so as to reduce the power consumption of the entire server, and there is no need to limit the distribution to the proportion of transactions or limit the ratio to natural numbers.

  In the above description, the allowable reciprocation time is determined from the measured reference reciprocation time. However, the allowable reciprocation time is acquired from the outside as a predetermined required specification and stored in the flash ROM 102 or the RAM 103. It may be

The load distribution apparatus 100 of the present invention can also be realized by a computer such as a normal PC.
Specifically, in each of the above embodiments, the program of the load distribution apparatus 100 has been described as being stored in advance in the flash ROM 102. However, the program of the flash ROM 102 may be installed in a computer to configure a computer capable of executing the functions of the respective units described above. The program is not limited to the flash ROM 102, and other computer readable recording media (for example, a flexible disk, a CD-ROM (Compact Disc-Read Only Memory), a DVD (Digital Versatile Disk), and an MO disc (Magneto Optical Disc) Of course, it may be stored in a) and distributed to a computer.

  The program may be stored in a disk device or the like included in a server device on a communication network such as the Internet, and may be downloaded to a computer, for example. Furthermore, the processing of the load distribution apparatus 100 described above can also be achieved by executing and executing the program while transferring it via the communication network.

1 to 5 servers 10 transaction number acquisition unit 20 minimum server number acquisition unit 21 reference round trip time acquisition unit 22 average round trip time acquisition unit 23 average CPU usage rate acquisition unit 30 maximum transaction number acquisition unit 40 server selection unit 50 distribution unit 100 ( 100 ') Load balancer 101 CPU
102 Flash ROM
103 RAM
104 Network I / F
210, 220, 230 clients

Claims (13)

A load balancer connected to a client and a plurality of servers via a network,
Measurement value acquisition means for acquiring a measurement value that is an indicator of the load on the plurality of servers;
Reference round trip time acquiring means for obtaining an average round trip time taken to receive response processing from the plurality of servers after transferring a processing request per unit time to the plurality of servers, as a reference round trip time;
Average round trip time obtaining means for obtaining the average round trip time by reducing the number of the plurality of servers;
The minimum number of servers obtained by specifying the number of servers whose average round trip time exceeds the predetermined allowable round trip time calculated from the reference round trip time and adding the number of one server to the specified number of servers as the minimum number of servers Acquisition means,
Processing performance with which the plurality of servers can process the processing request per unit time on the basis of the minimum number of servers obtained by the minimum server number obtaining means and the measured value obtained by the measured value obtaining means Processing performance value obtaining means for obtaining a value;
Selection means for selecting a server that responds to the processing request from the client among the plurality of servers based on the processing performance value obtained by the processing performance value acquisition means and the fluctuation of the measurement value;
Distribution means for distributing the processing request from the client to the server selected by the selection means;
A load distribution apparatus characterized by comprising: A load balancer connected to a client and a plurality of servers via a network,
Measurement value acquisition means for acquiring a measurement value that is an indicator of the load on the plurality of servers;
Average CPU utilization acquisition means for calculating the average CPU utilization from the CPU utilization of each server that processes response requests per unit time
A minimum number-of-servers acquisition unit that specifies the number of servers whose average CPU usage rate exceeds a predetermined upper limit rate, and adds the number of one server to the specified number of servers as the minimum number of servers;
Processing performance with which the plurality of servers can process the processing request per unit time on the basis of the minimum number of servers obtained by the minimum server number obtaining means and the measured value obtained by the measured value obtaining means Processing performance value obtaining means for obtaining a value;
Selection means for selecting a server that responds to the processing request from the client among the plurality of servers based on the processing performance value obtained by the processing performance value acquisition means and the fluctuation of the measurement value;
Distribution means for distributing the processing request from the client to the server selected by the selection means;
A load distribution apparatus characterized by comprising: The measurement value acquisition unit acquires the number of processing requests per unit time from the client as a measurement value that is an index of the load.
The load distribution device according to claim 1 or 2, characterized in that: The measurement value acquisition unit acquires, as a measurement value serving as an index of the load, the number of processes processed in response to the processing request per unit time from the client.
The load distribution device according to claim 1 or 2, characterized in that: The selection means is
The server which responds to the processing request from the client is selected among the plurality of servers so that there is a margin between the processing performance value determined by the processing performance value acquiring means and the fluctuating measurement value.
The load distribution device according to any one of claims 1 to 4, characterized in that: The distribution means is
When there are two or more servers selected by the selection means so as to allow the margin, each of the two or more weighting factors indicating a load distribution ratio so that the processing load includes a server lighter than other servers. Assigned to the server, and the processing request from the client is distributed based on the assigned weighting factor.
The load distribution device according to claim 5, wherein The distribution means is
The ratio of the weighting factor given to each server is dynamically changed according to the fluctuation of the measurement value, and the processing request from the client is distributed based on the changed ratio of the weighting factor.
7. The load distribution device according to claim 6, wherein: The distribution means is
Equalize the ratio of weighting factors given to each of the servers as the measurement value increases
The load distribution device according to claim 7, characterized in that: The processing performance value acquiring unit recalculates the processing performance value when there is an increase or decrease in the number of the plurality of servers.
The load distribution device according to any one of claims 1 to 8, characterized in that: A load balancing method performed by a load balancing apparatus connected to a client and a plurality of servers via a network, comprising:
A measurement value acquisition step of acquiring a measurement value indicative of load on the plurality of servers,
A reference round trip time obtaining step of obtaining an average round trip time taken to receive response processing from the plurality of servers after transferring a processing request per unit time to the plurality of servers, as a reference round trip time;
An average round trip time acquiring step of obtaining the average round trip time by reducing the number of the plurality of servers;
The minimum number of servers obtained by specifying the number of servers whose average round trip time exceeds the predetermined allowable round trip time calculated from the reference round trip time and adding the number of one server to the specified number of servers as the minimum number of servers Acquisition step,
Processing performance with which the plurality of servers can process the processing request per unit time on the basis of the minimum number of servers obtained in the minimum server number acquiring step and the measured value acquired in the measured value acquiring step Processing performance value acquisition step for obtaining a value;
A selection step of selecting a server to which the processing request from the client is to be subjected to response processing among the plurality of servers based on the processing performance value obtained in the processing performance value acquisition step and the fluctuation of the measurement value;
A distribution step of distributing the processing request from the client to the server selected in the selection step;
And a load distribution method characterized by comprising. A load balancing method performed by a load balancing apparatus connected to a client and a plurality of servers via a network, comprising:
A measurement value acquisition step of acquiring a measurement value indicative of load on the plurality of servers,
An average CPU utilization acquisition step of calculating an average CPU utilization from the CPU utilization of each server that processes the processing request per unit time in response;
A minimum number of servers acquiring step of specifying the number of servers having the average CPU usage rate exceeding a predetermined upper limit rate, and adding the number of one server to the number of identified servers as the minimum number of servers;
Processing performance with which the plurality of servers can process the processing request per unit time on the basis of the minimum number of servers obtained in the minimum server number acquiring step and the measured value acquired in the measured value acquiring step Processing performance value acquisition step for obtaining a value;
A selection step of selecting a server to which the processing request from the client is to be subjected to response processing among the plurality of servers based on the processing performance value obtained in the processing performance value acquisition step and the fluctuation of the measurement value;
A distribution step of distributing the processing request from the client to the server selected in the selection step;
And a load distribution method characterized by comprising. A computer connected via a network to a client and multiple servers,
Measurement value acquisition means for acquiring measurement values that are indicators of the load on the plurality of servers,
Reference round trip time acquiring means for obtaining an average round trip time taken to receive response processing from the plurality of servers after transferring a processing request per unit time to the plurality of servers, as a reference round trip time;
An average round trip time acquiring unit which obtains the average round trip time by reducing the number of the plurality of servers;
The minimum number of servers obtained by specifying the number of servers whose average round trip time exceeds the predetermined allowable round trip time calculated from the reference round trip time and adding the number of one server to the specified number of servers as the minimum number of servers Acquisition means,
Processing performance with which the plurality of servers can process the processing request per unit time on the basis of the minimum number of servers obtained by the minimum server number obtaining means and the measured value obtained by the measured value obtaining means Processing performance value acquisition means for determining the value,
Selecting means for selecting a server to which the processing request from the client is to be subjected to response processing among the plurality of servers based on the processing performance value obtained by the processing performance value acquiring means and the fluctuation of the measured value;
Distribution means for distributing the processing request from the client to the server selected by the selection means;
Program to function as. A computer connected via a network to a client and multiple servers,
Measurement value acquisition means for acquiring measurement values that are indicators of the load on the plurality of servers,
Average CPU utilization acquisition means for calculating the average CPU utilization from the CPU utilization of each server that processes response requests per unit time
A minimum number of servers acquiring unit for specifying the number of servers having the average CPU usage rate exceeding a predetermined upper limit rate and adding the number of one server to the number of identified servers as the minimum number of servers;
Processing performance with which the plurality of servers can process the processing request per unit time on the basis of the minimum number of servers obtained by the minimum server number obtaining means and the measured value obtained by the measured value obtaining means Processing performance value acquisition means for determining the value,
Selecting means for selecting a server to which the processing request from the client is to be subjected to response processing among the plurality of servers based on the processing performance value obtained by the processing performance value acquiring means and the fluctuation of the measured value;
Distribution means for distributing the processing request from the client to the server selected by the selection means;
Program to function as.

Images