JP2008059060A - Service linkage server and load balancing method - Google Patents

Abstract

[PROBLEMS] To achieve high scalability and flexibly cope with a rapid increase in demand.
In a web protocol load balancer 30 and a telecom protocol load balancer 40, when a signal is received, a distribution determination unit determines whether the signal includes a session identifier, and the signal includes a session identifier. If it is, it is determined whether the signal is for the servers 20-1 to 20-4. When the signal is for the servers 20-1 to 20-4, the IP address embedded in the session management unit is extracted from the session identifier, and the signal is transmitted to the IP address.
[Selection] Figure 1

Description

  The present invention relates to a service cooperation server and a load distribution method in a cooperation server for linking a web service and a telecom service.

  In the case of realizing high scalability in web services, VoIP (Voice over Internet Protocol) services, etc., it is common to perform load distribution by a load balancer (load distribution device) (for example, see Non-Patent Document 1).

  When applying a load balancer to a server that has common session data that needs to be referenced in multiple requests in HTTP (Hyper Text Transfer Protocol), SIP (Session Initiation Protocol), etc., It is necessary to distribute the request correctly to the server where the session data exists.

As a load balance method in consideration of a session, load balancing is generally performed by a load balancer based on a session identifier such as Cookie ID in HTTP or Call-ID in SIP. However, in a system in which both parties cooperate, a distribution method that is conscious of both sessions at the same time has not been studied.
By Tony Bourke, directed by Kimiaki Nabeshima, server load balancing technology, O'Reilly Japan, 2001, p185-188

  By the way, when linking a web service and a telecom service, it is effective to apply a service cooperation server that converts a complicated telecom protocol from an application into an easy-to-handle API. The service cooperation server is a system that handles both a web protocol such as HTTP and a telecom protocol such as H323 and SIP at the same time, and both requests may share one session data.

  In general, since the HTTP session identifier and the SIP session identifier are each set independently, in such a system, when load balancing is performed by a load balancer, distribution by the session identifier that is an existing method is performed. The method cannot cope. For this reason, it is necessary to realize a load balancing method that is conscious of sessions in both protocols.

  The present invention has been made in consideration of such circumstances, and the object thereof is to realize high scalability in the service cooperation server and to flexibly cope with a rapid increase in demand. It is to provide a service cooperation server and a load distribution method.

  In order to solve the above-described problem, the present invention provides a service cooperation server that distributes requests according to a plurality of protocols for services provided by a plurality of servers, and the request is generated when the requests according to the plurality of protocols are first generated. A session identifier generating means for generating a new session identifier from the session identifier of the server providing the service corresponding to the request, and requests by the plurality of protocols generated after the first occurrence of the request Determining means for determining whether the request is a request for any of the plurality of servers, and when the determination means determines that the request is a request for any of the plurality of servers, Request by protocol Extracting means for extracting a server identifier from a session identifier included in the client, and server specifying means for specifying a server that provides a service corresponding to the request based on the server identifier extracted by the extracting means. Is a service cooperation server characterized by

In order to solve the above-described problem, the present invention is a service cooperation server that distributes requests based on a plurality of protocols for services provided by a plurality of servers,
A service cooperation server comprising service data management means for centrally managing service data provided by the plurality of servers so that all services provided by the plurality of servers can be accessed. is there.

  In order to solve the above-described problem, the present invention provides a service cooperation server that distributes requests based on a plurality of protocols to services provided by a plurality of servers, and includes a session identifier included in the request and the plurality of servers. A distribution destination management storage unit that associates and stores a server identifier, and a server identifier acquisition unit that acquires a corresponding server identifier from the distribution destination management storage unit based on a session identifier included in the requests according to the plurality of protocols; And a server identification unit that identifies a server that provides a service corresponding to the request based on the server identifier acquired by the server identifier acquisition unit.

  According to the present invention, in the invention described above, the request is determined by the determining unit that determines whether the request based on the plurality of protocols is a request for any of the plurality of servers. A registration unit for registering the session identifier included in the request and the server identifier of the request transmission source in association with the distribution destination management storage unit when it is determined that the request is not for any of the servers. Features.

  The present invention relates to a load distribution method for distributing requests based on a plurality of protocols to services provided by a plurality of servers, and when a request based on the plurality of protocols is generated for the first time, the request identifier corresponds to the request. A new session identifier is generated from the server identifier of the server providing the service to be executed, and whether or not the request by the plurality of protocols generated after the first generation of the request is a request to any of the plurality of servers If it is determined that the request is for one of the plurality of servers, a server identifier is extracted from a session identifier included in the request according to the plurality of protocols, and the request is based on the extracted server identifier. Corresponding to A load distribution method characterized by identifying a server providing-bis.

  The present invention is a load distribution method for distributing requests based on a plurality of protocols to services provided by a plurality of servers, wherein the plurality of servers are accessible so that all services provided by the plurality of servers can be accessed. The load distribution method is characterized in that the service data provided by is managed centrally.

  The present invention is a load distribution method for distributing requests based on a plurality of protocols to services provided by a plurality of servers, and stores session identifiers included in the requests and server identifiers of the plurality of servers in association with each other. A load characterized by acquiring a server identifier corresponding to a session identifier included in a request according to the plurality of protocols, and identifying a server that provides a service corresponding to the request based on the acquired server identifier Dispersion method.

  According to the present invention, when a request according to a plurality of protocols is generated for the first time, a new session identifier is generated from the session identifier of the request and the server identifier of the server providing the service corresponding to the request, and the first generation It is determined whether or not a request according to the plurality of protocols generated after the time is a request for any of the plurality of servers, and when it is determined that the request is for any of the plurality of servers, A server identifier is extracted from the session identifier included in the request, and a server that provides a service corresponding to the request is specified based on the extracted server identifier. Therefore, it is possible to obtain an advantage that a server that provides a service corresponding to the request can be easily specified from requests according to a plurality of protocols.

  In addition, according to the present invention, service data provided by a plurality of servers is centrally managed so that all services provided by the plurality of servers can be accessed. Therefore, there is an advantage that the service data can be distributed without being aware of the management position.

  Further, according to the present invention, the session identifier included in the request and the server identifiers of the plurality of servers are stored in association with each other, and the server identifier corresponding to the session identifier included in the request according to the plurality of protocols is acquired, Based on the acquired server identifier, a server that provides a service corresponding to the request is specified. Therefore, there is an advantage that a server that provides a service corresponding to the request can be easily specified from a request based on a plurality of protocols without newly generating a session identifier.

  Further, according to the present invention, it is determined whether or not a request based on a plurality of protocols is a request for any of a plurality of servers, and if it is determined that the request is not a request for any of a plurality of servers, The session identifier included and the server identifier of the request transmission source are stored in association with each other. Therefore, the association between the session identifier included in the request and the server identifiers of a plurality of servers can be easily updated, and the server that provides the service corresponding to the request can always be easily identified with the latest information. The advantage is obtained.

  Hereinafter, a load balancing method according to an embodiment of the present invention will be described with reference to the drawings.

A. First Embodiment FIG. 1 is a block diagram showing a configuration of a service cooperation server to which a service based on an HTTP protocol and a SIP protocol cooperates, to which a load distribution method according to a first embodiment of the present invention is applied. In FIG. 1, a web-telecom cooperation server 10 is a server that processes HTTP and SIP signals, and includes a plurality of servers 20-1 to 20-4, a web protocol load balancer 30, and a telecom protocol load balancer 40. ing.

  Each of the servers 20-1 to 20-4 provides a service according to the request. The web protocol load balancer 30 is connected to the Internet 45 using HTTP, and the telecom protocol load balancer 40 is connected to a telecom network 46 using SIP. Each of the web protocol load balancer 30 and the telecom protocol load balancer 40 distributes both HTTP and SIP requests transmitted from the client 50 to one of the servers 20-1 to 20-4 where service data exists. .

  The client 50 sends both HTTP and SIP requests to the web protocol load balancer 30 and the telecom protocol load balancer 40 via the Internet 45 and the telecom network 46, respectively.

  Next, FIG. 2 is a block diagram showing functional configurations of the servers 20-1 to 20-4, the web protocol load balancer 30, and the telecom protocol load balancer 40. In FIG. 2, the servers 20-1 to 20-4 each include a service data management unit 60 and a session management unit 70. The service data management unit 60 generates service data when a service is started in the server. The session management unit 70 issues a session identifier to a request that activates a service or a session generated by the activated service. In subsequent requests, access to the service data is performed by designating the session identifier.

  In the above configuration, the web protocol load balancer 30 and the telecom protocol load balancer 40 have a service data management unit 60 in which service data related to the session exists for a request for which a session identifier is specified. It is necessary to distribute request signals to i (i = 1, 2, 3, 4).

  Therefore, in the first embodiment, when the session management unit 70 issues a session identifier, the session identifier and the server 20-i (i = 1, 2, 3, 4) where the session is generated are generated. A new session identifier is generated from the IP address. As a method for generating a new session identifier, a new session identifier may be generated by concatenating an IP address to the session identifier. Alternatively, a new session identifier may be generated from the session identifier and the IP address by a hash operation or the like. May be. At the time of distribution by the web protocol load balancer 30 and the telecom protocol load balancer 40, the IP address embedded in the session identifier is extracted and signal transmission is performed to the server 20-i corresponding to the IP address. That's fine.

  Next, FIG. 3 is a flowchart for explaining the operations of the web protocol load balancer 30 and the telecom protocol load balancer 40 to which the load distribution method according to the first embodiment is applied. In the web protocol load balancer 30 and the telecom protocol load balancer 40, the distribution determination unit 80 determines whether the session identifier is included in the signal when the signal is received (step S10), and the session identifier is included in the signal. If so, it is determined whether or not the signal is for the servers 20-1 to 20-4 (step S11). If the signal is for the servers 20-1 to 20-4, the IP address is extracted from the session identifier (step S12), and the signal is transmitted to the IP address.

  For example, in a certain service, when a new session is generated between clients by SIP, as shown in FIG. 4, the server 20-i (i = 1, 2, 3, 3) is used as the Call-ID used as the SIP session identifier. 4) Concatenate server identifiers such as IP addresses and use the values as new Call-IDs. Then, when the SIP signal is distributed by the telecom protocol load balancer 40, the server 20-i that is the distribution destination is determined by taking out the IP address linked from the designated Call-ID.

  Also in the case of HTTP, a server identifier such as the IP address of the server 20-i (i = 1, 2, 3, 4) is concatenated with the HTTP session identifier to form a new session identifier, thereby making the HTTP When the signal is distributed, it is possible to determine the server 20-i that is the distribution destination.

B. Second Embodiment Next, a second embodiment of the present invention will be described. In the second embodiment, the service data management unit 60 can be accessed from all the servers 20-1 to 20-4.

  FIG. 5 is a block diagram showing the configuration of the servers 20-1 to 20-4 according to the second embodiment. In FIG. 5, all the servers 20-1 to 20-4 are provided with a service data centralized management unit 61 that centrally manages service data so that all service data can be accessed. In addition to the provision of the service data centralized management unit 61 in each of the servers 20-1 to 20-4, a common database that can be referred to from all the servers 20-1 to 20-4 may be separately provided.

  When a request distributed to each of the servers 20-1 to 20-4 accesses service data, the service data management unit 60 is accessed to perform processing. In this case, the distribution destination determination unit 80 shown in FIG. 2 does not need to perform distribution particularly in consideration of the session identifier, and can use a known distribution method such as a round robin method or a weighted round robin method. .

  As a result, the web protocol load balancer 30 and the telecom protocol load balancer 40 can distribute the service data without regard to the management position of the service data.

C. Third Embodiment Next, a third embodiment of the present invention will be described. In the third embodiment, the service data management unit 60 and the session management unit 70 distribute request signals without performing special processing.

  FIG. 6 is a block diagram showing a configuration of a web protocol load balancer 30 and a telecom protocol load balancer 40 to which the load distribution method according to the third embodiment is applied. In FIG. 6, the web protocol load balancer 30 and the telecom protocol load balancer 40 include a distribution destination management table 81. As shown in FIG. 7, the distribution destination management table 81 includes servers 20-1 to 20- that provide session identifiers S-ID1, S-ID2, S-ID3,... And service data corresponding to the session identifiers. 4 IP addresses IP1, IP2, IP3,...

  The distribution destination determination unit 80 receives a request signal, and if a session identifier is present in the signal, if it is a signal for the server 20-i (i = 1, 2, 3, 4), The IP address of the distribution destination server corresponding to the session identifier is acquired from the distribution destination management table 81, and a signal is transmitted to the server 20-i with the IP address. Further, when the signal is not a signal for the servers 20-1 to 20-4, that is, when the signal is for the client 50, the distribution destination determination unit 80 stores a session identifier and a signal transmission source (server 20-2) in the distribution destination management table 81. i) is registered with the IP address and a signal is transmitted to the destination address.

  Next, FIG. 8 is a flowchart for explaining the operation of the third embodiment. In the web protocol load balancer 30 and the telecom protocol load balancer 40, the distribution determination unit 80 determines whether the session identifier is included in the signal when receiving the signal (step S20), and the signal includes the session identifier. If so, it is determined whether or not the signal is for the servers 20-1 to 20-4 (step S21). If it is a signal for the servers 20-1 to 20-4, the IP address of the distribution destination server corresponding to the session identifier is acquired from the distribution destination management table 81 (step S22), and a signal is sent to the IP address. Send.

  On the other hand, if it is not a signal for the servers 20-1 to 20-4, the session identifier and the IP address of the signal transmission source (server 20-i) are registered in the distribution destination management table 81 (step S23), and the transmission destination address. Send a signal to

  Here, FIG. 9 is a sequence diagram for explaining an operation example of the third embodiment. The sequence shown in FIG. 9 is an operation example when a SIP session is controlled by HTTP, and client A uses HTTP, and clients B and C communicate using SIP. The HTTP server 20-A corresponds to one of the servers 20-i (i = 1 to 4) shown in FIG. 1, and provides a service using HTTP. The SIP server 20-B corresponds to one of the servers 20-i (i = 1 to 4) shown in FIG. 1 and provides a service using SIP.

  First, when there is a call request (request) from the HTTP client A to the HTTP server 20-A (SS1), the HTTP server 20-A connects between the SIP clients B and C specified by the HTTP request data. A session start request is made to the SIP server 20-B so as to establish a session (SS2).

  Next, the SIP server 20-B notifies the SIP server identifier to the HTTP server 20-A (SS3). The HTTP server 20-A that has received the response from the SIP server 20-B newly issues an HTTP session identifier, stores the correspondence between the HTTP session identifier and the SIP session identifier as HTTP session data, and then stores it in the HTTP client A. On the other hand, an HTTP session identifier is designated as the identifier of the HTTP session, and a session establishment response is transmitted (SS4).

  At this time, since the session establishment response is not for the server, as described above, in the web protocol load balancer 30, the HTTP session identifier and the IP of the signal transmission source (HTTP server 20-A) are stored in the distribution destination management table 81. The address is registered (SS5) and transmitted to the client A of HTTP (SS6).

  In addition, the SIP server 20-B that has received the request issues a new SIP session identifier (Call-ID), and first designates the SIP session identifier and notifies the client B of the start of the session (SS7). At this time, since the session start notification is not for the server, as described above, in the telecom protocol load balancer 40, the SIP session identifier and the IP of the signal transmission source (SIP server 20-B) are stored in the distribution destination management table 81. The address is registered (SS8) and transmitted to the SIP client B (SS9). Client B specifies the SIP session identifier and returns a session start response (SS10).

  In the telecom protocol load balancer 40, when a session start response from the client B is received, it is a signal to the server, so the IP address of the distribution destination server corresponding to the SIP session identifier from the distribution destination management table 81, in this case, the SIP server The 20-B IP address is acquired (SS11), and a session start response is transmitted to the IP address (SS12).

  Similarly, the SIP server 20-B designates the newly generated SIP session identifier and notifies the client C of the start of the session (SS13). At this time, since the session start notification is not for the server, as described above, in the telecom protocol load balancer 40, the SIP session identifier and the IP of the signal transmission source (SIP server 20-B) are stored in the distribution destination management table 81. The address is registered (SS14) and transmitted to the SIP client C (SS15). The client C specifies the SIP session identifier and returns a session start response (SS16).

  In the telecom protocol load balancer 40, when a session start response is received from the client C, it is a signal to the server, so the IP address of the distribution destination server corresponding to the SIP session identifier from the distribution destination management table 81, in this case, the SIP server The 20-B IP address is acquired (SS17), and a session start response is transmitted to the IP address (SS18).

  In order to confirm whether or not the SIP session is established in the HTTP server 20-A, when a request for inquiring the SIP session state by specifying the SIP session identifier is transmitted to the HTTP server 20-A, the SIP server The server 20-B is inquired about the status, and if the SIP client C is before responding, a value indicating "waiting for connection" is returned, and if the SIP client C is responding, a value indicating the status of "connecting" is returned. come. The client A of HTTP confirms the above value and determines whether or not it is connected.

  After the session is established, when the end of the SIP session or the SIP session state information is referred from the HTTP client A, the HTTP session identifier is specified and a request is transmitted to the HTTP server 20-A. For example, at the end of the SIP session, an HTTP session identifier is designated and an end request is transmitted to the HTTP server 20-A (SS19).

  When receiving a termination request from the client A, the web-system protocol load balancer 30 sends a signal to the server, so the IP address of the distribution destination server corresponding to the HTTP session identifier from the distribution destination management table 81, in this case, the HTTP server 20 -Obtain the IP address of A (SS20), and send a termination request to the IP address (SS21). The HTTP server 20-A extracts the SIP session identifier from the session data, and makes a session end request to the SIP server 20-B (SS22).

  The SIP server 20-B designates the SIP session identifier and transmits a session end notification to the client B (SS23). At this time, since the session end notification is not for the server, as described above, in the web-system protocol load balancer 30, the SIP session identifier and the IP of the signal transmission source (SIP server 20-B) are stored in the distribution destination management table 81. The address is registered (SS24) and transmitted to the SIP client B (SS25). From the client B, a session end response is returned by designating the SIP session identifier (SS26).

  In the telecom protocol load balancer 40, when a session end response from the client B is received, it is a signal to the server, so the IP address of the distribution destination server corresponding to the SIP session identifier from the distribution destination management table 81, in this case, the SIP server The 20-B IP address is acquired (SS27), and a session end response is transmitted to the IP address (SS28).

  Similarly, the SIP server 20-B transmits a session end notification designated as a SIP session identifier to the client C (SS29). At this time, since the session end notification is not to the server, as described above, in the telecom protocol load balancer 40, the SIP session identifier and the IP of the signal transmission source (SIP server 20-B) are stored in the distribution destination management table 81. The address is registered (SS30) and transmitted to the SIP client C (SS31). The client C returns a session end response by designating the SIP session identifier (SS32).

  In the telecom protocol load balancer 40, when a session end response is received from the client C, it is a signal to the server, so the IP address of the distribution destination server corresponding to the SIP session identifier from the distribution destination management table 81, in this case, the SIP server The 20-B IP address is acquired (SS33), and a session end response is transmitted to the IP address (SS34). The SIP server 20-B transmits the session end response to the HTTP server 20-A (SS35), and the HTTP server 20-A transmits the session end response to the HTTP client A (SS36).

  In order to realize the operation of the sequence shown in FIG. 8 with the configuration of the first embodiment, when issuing the session identifier in the HTTP server 20-A and the SIP server 20-B, the IP address, etc. The server ID including the server identifier is issued. In the configuration of the second embodiment, the HTTP server 20-A and the SIP server 20-B are configured to include the service data centralized management unit 61 described above.

  According to the first to third embodiments described above, high scalability can be realized in the service cooperation server, and a system that can flexibly cope with a rapid increase in demand can be constructed.

It is a block diagram which shows the structure of the service cooperation server to which the service by the HTTP protocol and SIP protocol which apply the load distribution system by 1st Embodiment of this invention cooperates. It is a block diagram which shows the function structure of the servers 20-1 to 20-4, the web type | system | group protocol load balancer 30, and the telecom type | system | group protocol load balancer 40. FIG. It is a flowchart for demonstrating the operation | movement of this 1st Embodiment. It is a conceptual diagram which shows the example of embedding IP address to a session identifier. It is a block diagram which shows the structure of the servers 20-1 to 20-4 by this 2nd Embodiment. It is a block diagram which shows the structure of the web type | system | group protocol load balancer 30 and the telecom type | system | group protocol load balancer 40 to which the load distribution system by this 3rd Embodiment is applied. It is a conceptual diagram which shows the data structure of the distribution destination management table 81 by this 3rd Embodiment. It is a flowchart for demonstrating the operation | movement of this 3rd Embodiment. It is a sequence diagram for demonstrating the operation example of this 3rd Embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Web-telecom service cooperation server 20-1-20-4 server 30 Web system protocol load balancer 40 Telecom system protocol load balancer 45 Internet 46 Telecom network 50 Client 60 Service data management part 61 Service data central management part (Service data management means )
70 Session management unit (session identifier generating means)
80 Distribution destination determination unit (extraction means, server identification means, determination means, registration means)
81 Distribution destination management table (distribution destination management storage means)
A HTTP client B, C SIP client 20-A HTTP server 20-B SIP server

Claims (7)

A service linkage server that distributes requests based on multiple protocols for services provided by multiple servers,
Session identifier generating means for generating a new session identifier from a session identifier of the request and a server identifier of a server providing a service corresponding to the request at the first occurrence of a request according to the plurality of protocols;
Determining means for determining whether a request according to the plurality of protocols generated after the first occurrence of the request is a request for any of the plurality of servers;
An extracting unit that extracts a server identifier from a session identifier included in a request according to the plurality of protocols when the determination unit determines that the request is a request to any of the plurality of servers;
And a server identification unit that identifies a server that provides a service corresponding to the request based on the server identifier extracted by the extraction unit. A service linkage server that distributes requests based on multiple protocols for services provided by multiple servers,
A service cooperation server comprising service data management means for centrally managing service data provided by the plurality of servers so that all services provided by the plurality of servers can be accessed. A service linkage server that distributes requests based on multiple protocols for services provided by multiple servers,
A distribution destination management storage means for storing a session identifier included in the request and server identifiers of the plurality of servers in association with each other;
Server identifier acquisition means for acquiring a corresponding server identifier from the distribution destination management storage means based on a session identifier included in the requests by the plurality of protocols;
And a server identification unit that identifies a server that provides a service corresponding to the request based on the server identifier acquired by the server identifier acquisition unit. Determining means for determining whether a request according to the plurality of protocols is a request to any of the plurality of servers;
When the determination unit determines that the request is not a request to any of the plurality of servers, the distribution destination management storage unit associates the session identifier included in the request with the server identifier of the request transmission source. The service cooperation server according to claim 3, further comprising registration means for registering with the service cooperation server. A load distribution method for distributing requests by a plurality of protocols for services provided by a plurality of servers,
Generating a new session identifier from a session identifier of the request and a server identifier of a server providing a service corresponding to the request at the first occurrence of the request by the plurality of protocols;
Determining whether a request according to the plurality of protocols generated after the first occurrence of the request is a request to any of the plurality of servers;
When it is determined that the request is for one of the plurality of servers, a server identifier is extracted from a session identifier included in the request by the plurality of protocols, and a service corresponding to the request is determined based on the extracted server identifier. A load distribution method characterized by specifying a server to be provided. A load distribution method for distributing requests by a plurality of protocols for services provided by a plurality of servers,
A load distribution method characterized in that service data provided by the plurality of servers is centrally managed so that all services provided by the plurality of servers can be accessed. A load distribution method for distributing requests by a plurality of protocols for services provided by a plurality of servers,
The session identifier included in the request and the server identifiers of the plurality of servers are stored in association with each other, the server identifier corresponding to the session identifier included in the request according to the plurality of protocols is acquired, and the acquired server identifier is A load distribution method comprising: identifying a server that provides a service corresponding to the request based on the request.

Images