JP4028627B2 - Client server system and communication management method for client server system - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a client server system related to a system in which a communication device of one computer is replaced with a communication device of another computer in a LAN (local area network) environment. And client server system communication management method About.
[0002]
[Prior art]
In conventional techniques such as computer cluster and hot standby, when a failure occurs in a computer and the processing cannot be continued, an alternative computer takes over the processing.
[0003]
There are various types of clusters. As an example, taking over of cluster processing will be described with reference to FIG. There are a primary server 101 and a standby server 110, and a cluster control SW (software) that takes over processing is operating. Normally, each server performs a different process. The cluster control SW periodically checks whether the other party is operating correctly using the control communication link 120.
[0004]
When a failure occurs in the primary server 101 and the cluster control SW of the primary server 101 detects this, the primary server 101 stops the process and notifies the cluster control SW of the standby server 110. Alternatively, a failure of the primary server 101 is detected when the cluster control SW of the standby server 110 periodically inspects the primary server 101. In any of the above, the cluster control SW of the standby server 110 can detect that the main server 101 has stopped processing.
[0005]
Next, the standby server 110 starts to take over the processing that has been executed by the primary server 101. The process to be taken over needs to be defined in advance. The following two points are necessary to take over the processing.
(1) Prepare an environment that produces the same results regardless of the computer
For example, if the primary server provides a database service, the standby server must be able to access the database file even if the primary server has stopped processing in order to take over the service on the standby server. become.
(2) Switching communication destination
When communicating with other computers via the network, the communication partner is switched from the primary server to the standby server so that the partner computer is not aware.
[0006]
Although a method for realizing the above (1) has been proposed, description thereof is omitted here. Here, switching of the communication destination (2) will be described.
Services provided by the primary server 101 to other computers exchange requests and results through the network. Other computers that communicate with the primary server 101 continue to communicate with the primary server 101 even if a failure occurs in the primary server 101 and the primary server 101 stops processing.
[0007]
In order to take over these communications by the standby server 110, the standby server 110 receives a packet addressed to the address of the primary server 101, and transmits a packet from the address of the primary server 101. Since TCP / IP can have a plurality of IP addresses, this function is used.
[0008]
Normally, standby server 110 receives only its own IP address. After the primary server 101 stops processing, the cluster control SW of the standby server 110 sets the TCP / IP software so that the IP address of the primary server 101 can also be received. Thus, the packet of the primary server 101 can be received, and other computers do not have to make a change due to the stop of the primary server 101.
[0009]
If the primary server 101 returns, the standby server 110 stops the service that it has taken over and stops receiving the packet of the primary server 101 for communication. JP-A-8-212095 exists as the above technique.
[0010]
[Problems to be solved by the invention]
However, in the above-described prior art, if there is a failure in the communication device for the LAN 121 of the primary server, the standby server 110 takes over the communication addressed to the primary server 101, so that the standby server 110 is connected to the LAN 121 of the primary server 101. There is a problem that the change must be notified so as to receive all the logical addresses for each transport (TCP / IP, NETBEUI, etc.) instead of the physical address of the network.
[0011]
The present invention has been made in view of the above circumstances, and has solved the above problems, and enables a standby server to receive the physical address of the LAN of the primary server. And client server system communication management method The purpose is to provide.
[0012]
[Means for Solving the Problems]
Up To achieve the purpose, The client server system of the present invention In a client server system having a primary server and a standby server, a dedicated communication link for connecting the primary server and the standby server; It is provided between the main server and the standby server and the LAN, has a plurality of ports to which the LAN interface of the computer is connected, and stores the physical address of the LAN interface of each port destination. A LAN switch that transfers packets addressed to the stored physical address taken in from the LAN to only the LAN interface of the corresponding port destination among the plurality of ports, and the primary server includes: When a failure that makes it impossible to execute communication processing via the LAN using the local server's LAN interface occurs, the physical address of the local server's LAN interface is notified to the standby server via the dedicated communication link. , Acting as a proxy for sending and receiving packets for communication processing via the LAN via the dedicated communication link Means for requesting the standby server via a dedicated communication link, the standby server acting as a proxy for sending and receiving packets for communication processing via the LAN via the dedicated communication link The main server is set to a mode for receiving all packets from the LAN, and the main interface notified via the dedicated communication link is set. A packet having the physical address of the local server's LAN interface as the source address is sent to the LAN using the local server's LAN interface, and the packet addressed to the physical address of the primary server's LAN interface is transmitted to the local server's LAN interface. Means for causing the LAN switch to learn to forward to It is characterized by that.
[0013]
Ma The present invention The client server system communication management method includes a main server, a standby server, a dedicated communication link connecting the main server and the standby server, the main server, and the standby server. And a plurality of ports connected to the computer's LAN interface, storing the physical address of the LAN interface of each port destination, and storing that taken in from the LAN A client server system communication management method comprising: a LAN switch that transfers a packet addressed to a physical address to only a LAN interface of a corresponding port destination among the plurality of ports. When a failure occurs that makes it impossible to execute communication processing via the LAN using the local server's LAN interface. Notifying the standby server via the dedicated communication link of the physical address of the local server's LAN interface, and acting as a proxy for sending and receiving packets for communication processing via the LAN via the dedicated communication link Requesting the standby server via the dedicated communication link, the standby server using the dedicated communication link as a proxy for sending and receiving packets for communication processing via the LAN via the dedicated communication link. When a request is received from the primary server via the dedicated server link, a packet having a physical address of the LAN interface of the primary server notified via the dedicated communication link as a source address is transmitted using the LAN interface of the local server. The packet sent to the LAN and addressed to the physical address of the LAN interface of the primary server Wherein to learn the LAN switch to forward to the LAN interface, It is characterized by that.
[0020]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings. In this embodiment, the primary server and the standby server are named for ease of explanation, but there is no essential difference between the primary server and the standby server, and separate processing is performed independently. Is possible.
(First embodiment)
FIG. 1 is a block diagram showing a system configuration in an embodiment of the present invention. Here, each component of the main server 1 will be mainly described, but the standby server 10 may have the same configuration.
[0021]
In the servers 1 and 10, the transports 5 and 14 serve to pass packets received via the LAN interfaces 2 and 11 and the LAN drivers 3 and 12 to the server processes 7 and 16. Here, for ease of explanation, the transport is assumed to be TCP / IP.
[0022]
The LAN interfaces 2 and 11 are devices that transmit and receive packets to and from the LAN 21. When a packet addressed to a physical address designated in advance is transmitted on the LAN 21, the packet is received and notified to the LAN drivers 3 and 12. Further, the packet delivered from the LAN drivers 3 and 12 is transmitted on the LAN 21. Here, in order to make the explanation easy to understand, it is assumed that the LAN is Ethernet.
[0023]
The LAN drivers 3 and 12 control the LAN interfaces 2 and 11. Packets received by the LAN interfaces 2 and 11 are processed into data that can be used by the transports 5 and 14 and transferred to the transports 5 and 14 and the communication management units 4 and 13. Further, a packet to be transmitted to the LAN 21 is created from the data passed from the transports 5 and 14 and is transmitted to the LAN 21 through the LAN interfaces 2 and 11. Further, when the LAN interfaces 2 and 11 are used, the failure is detected and notified to the communication management units 4 and 13.
[0024]
The communication link 20 is a dedicated link for performing communication between the servers 1 and 10.
The server processes 7 and 16 provide services that can be used from other computers on the network. For example, a database service is performed. A plurality of server processes 7 and 16 can exist, and the contents of the service may differ between the primary server 1 and the standby server 10.
[0025]
The client 22 is a computer that uses the services of the primary server 1 and the standby server 10 via the LAN. There can be multiple clients.
The communication management units 4 and 13 hold the failure state of the primary server 1 and the standby server 10 and control processing when the failure occurs and when it is normal. There are four states to be held: both servers are normal, a failure has occurred in the own server, a failure has occurred in another server, and a failure has occurred in both servers. There are various types of failures and detections are various, but in the embodiment of the present invention, only the case where communication processing through the LAN 21 cannot be performed is concerned. Therefore, only communication processing through the LAN 21 cannot be performed, and only when all other processing can be performed is treated as a failure.
[0026]
For example, this failure occurs when the LAN interface 2 becomes unusable. A failure of the LAN interface 2 can be detected when the LAN driver 3 transmits / receives a packet.
[0027]
Further, the cluster control unit 6 can periodically take a method of inspecting the LAN interface 2 through the LAN driver 3.
Another possible method is to check whether the server is operating correctly by transmitting data from the standby server 10 to the main server 1 via the LAN 21 and checking the contents thereof.
[0028]
There are also various methods for diagnosing recovery from failures. For example, a packet is transmitted from the LAN interface 11 of the standby server 10 in which no failure has occurred to the LAN interface 2 to periodically check whether it can be received. Alternatively, it is conceivable to periodically perform a process for checking whether or not the main server 1 has recovered from the failure.
[0029]
In the present invention, detection of failure occurrence and recovery is performed by the hardware of the primary server 1 or software such as the LAN driver 3 to notify the communication management unit 4 or by the communication management unit 4 itself. To do.
[0030]
FIG. 4 shows a state change of the communication management unit 4. When the communication management unit 4 detects a failure of the primary server 1 (own server), the communication management unit 4 enters a failure state in the own server and notifies the communication management unit 13 of the standby server 20 through the communication link 20.
[0031]
Upon receiving the notification, the communication management unit 13 of the standby server 20 enters a state where a failure has occurred in another server (RB).
Further, when recovering from the failure, both servers are made normal, and the communication management unit 13 is notified in the same manner as the failure notification.
[0032]
The communication management unit 13 receives the notification and changes to a normal state of both servers (RA).
If both servers are out of order, it is possible only when a plurality of LAN interfaces are provided as will be described later. Here, for the sake of explanation, only one LAN interface is used, and when there are a plurality of LAN interfaces, only the difference will be described later. Therefore, if both servers fail, communication cannot be taken over.
[0033]
The communication management unit 4 is a part for switching communication at the time of failure, which is a feature of the present invention. The communication management unit 4 is located between the LAN driver that controls communication to the actual LAN and the transport, and controls data exchange between them. Further, when a failure occurs, communication with the communication management unit 13 is performed through the communication link 20.
[0034]
The communication management unit 4 can detect a failure of the LAN interface 2 through the LAN driver 3. Further, the address of the received packet from the LAN passed from the LAN driver 3 can be inspected. Further, since the LAN driver 3 is a function only for transmitting the passed packet to the LAN, the address of the transmitted packet Can also be changed.
[0035]
FIG. 5 is a flowchart showing the operation processing procedure of the communication management unit 4.
In steps 60 and 62, it is determined whether it is failure notification or transmission or reception processing. From step 66, it is a reception process. It also handles both notifications from other servers' communication management units and from its own server, and data transmission / reception. For example, when data is sent to another server in step 70, the communication management unit of the other server enters the process of step 62 as a transmission process, and executes the flow of steps 64 and 65 as the other server failure.
[0036]
Next, the communication method will be described separately for the normal time and the failure time.
The normal process will be described with reference to the conceptual diagram showing the normal communication process in FIG.
Both of the communication management units 4 and 13 are in a normal state, and each server communicates using a built-in LAN interface.
[0037]
Data transmitted by the server process is processed by the transport 5, the communication management unit 4, and the LAN driver 3 and transmitted to the LAN 21 through the LAN interface 2. A packet that arrives from the LAN 21 to the LAN interface 2 follows the reverse path and reaches the server process 7.
[0038]
The communication management unit 4 is in a normal state of both servers, and only relays data transmitted and received between the transport 5 and the LAN driver 3.
Next, the process at the time of failure will be described using the case where a failure has occurred in the primary server 1 as an example using the conceptual diagram showing the communication processing at the time of failure in FIG. The communication management unit 4 detects the failure and sets the state as the own server failure.
[0039]
Next, the communication management unit 13 is notified of the failure occurrence through the communication link 20. The notification is sent together with the physical address that the other party wants to take over. The communication management unit 13 receives the failure notification and enters into a state of failure of other servers.
[0040]
The communication management unit 4 in a state of failure of its own server transmits data to be transmitted to the LAN 21 transmitted from the server process 7 or the like to the communication management unit 13 through the communication link 20. Conversely, the reception data sent from the communication management unit 13 is passed to the server process 7.
[0041]
The communication management unit 13 in the other server failure state transmits / receives a packet arriving from the LAN 21 to the physical address of the LAN interface 11 and a packet issued by the server process 16 in the same manner as in normal times. When a failure notification is received from the primary server 1, the LAN interface 12 is used to change the setting so that the LAN interface 11 also receives packets addressed to the physical address of the LAN interface 2.
[0042]
A packet transmitted from the communication management unit 4 through the communication link 20 is transmitted from the LAN interface 11 as a packet transmitted from the physical address of the LAN interface 2. Specifically, a packet in which the physical address of the LAN interface 2 is included in the address of the packet sender is sent to the LAN 21.
[0043]
The communication management unit 13 can determine whether the packet addressed to the LAN interface 2 received by the LAN interface 11 is addressed to the primary server 1 or the standby server 10 by looking at the packet destination. The server addressed to the standby server 10 is transferred to the transport 14 as it is in the normal processing. Those addressed to the main server 1 are transmitted to the communication management unit 4 through the communication link 20.
[0044]
A case will be described in which the failure of the primary server 1 disappears and the primary server 1 can resume communication.
It is assumed that the communication management unit 4 detects that a failure has disappeared. The communication link 20 is used to notify the communication management unit 13 so that both communication management units are in a normal state on both servers.
[0045]
The subsequent data transmission / reception is performed by each server using its own LAN interface as shown in FIG. Even when there are a plurality of LAN interfaces, it is possible to cope with them. However, the standby server 10 must be connected to the LAN to which the main server 1 is connected.
[0046]
If the communication management unit 4 sends a failure notification to the communication management unit 13 of the standby server 10 together with information indicating the LAN interface connected to which LAN has failed, the method described so far Thus, the standby server 10 can take over transmission / reception of the main server 1. In this case, what is necessary is just to hold | maintain the information of which LAN communication is impossible and to determine operation | movement.
[0047]
Depending on the type of LAN interface, only one physical address received from the LAN 21 can be set. In the above description, it is assumed that two physical addresses can be received in total for the primary server 1 and the standby server 10. In this case, the LAN driver 12 of the standby server 10 changes the LAN interface to a mode for receiving all packets transmitted to the LAN 21.
[0048]
The LAN driver 12 relays all received packets to the communication management unit 13. The communication management unit 13 can solve this problem by performing a process of selecting only packets addressed to the standby server 10 and packets addressed to the primary server 1 from the received packets.
[0049]
As described above, according to the present embodiment, when a failure occurs in the primary server 1 and communication processing cannot be performed, the communication processing can be continued through the LAN interface 11 of the standby server 10.
[0050]
Since the physical address of the LAN 21 is used instead of the logical address used by the transport, it is not necessary to notify other computers on the LAN 21 and change the combination of the logical address and the physical address.
[0051]
Further, depending on the client 22, there is a case where a combination of a logical address and a physical address is statically held therein and cannot be changed by a notification from the standby server 10. In this embodiment, since the standby server 10 receives data instead of the main server 1 without changing this combination of clients, it is possible to cope with such a client on the LAN 21. It is.
(Second embodiment)
In the first embodiment described above, there are specifications that can receive only one physical address depending on the type of LAN interface. When the standby server having such a LAN interface tries to receive a packet addressed to the physical address of the primary server as in the first embodiment, as described above, it receives all the packets, The communication management unit needs to perform processing for selecting packets addressed to the physical address of the LAN interface of the standby server and the physical address of the LAN interface of the primary server.
[0052]
In this case, since the LAN interface receives all packets on the LAN, there is a problem that the processing becomes very large, and the communication management unit of the standby server also has a large amount of processing.
[0053]
Therefore, as shown in FIG. 6, a system configuration using a commonly used LAN switch 30 will be described as a second embodiment. In FIG. 6, the same components as those in the first embodiment are denoted by the same reference numerals as those in FIG. 1, and the description thereof is omitted.
[0054]
Here, the LAN switch 30 is an Ethernet switch used in Ethernet. The Ethernet switch has a plurality of ports to which the LAN interface of the computer is connected. At the end of the port, not only one computer but also a plurality of computers connected by a hub may be used.
[0055]
The Ethernet switch scans a packet output from a port and stores the physical address of the LAN interface of the computer connected to each port for each port.
[0056]
When the Ethernet switch receives a packet addressed to a physical address stored as being connected to each port, the packet is transmitted only to that port and not transmitted to other packets.
[0057]
In a normal hub or the like, packets are always transmitted to all ports, but extra traffic can be suppressed by transmitting packets only to the ports connected as described above.
[0058]
The same operation as that of the first embodiment is performed at both the normal time and the time of failure. However, when the communication management unit 13 starts an operation to take over the communication of the primary server 1, it broadcasts a packet with the physical address of the LAN interface 2 of the primary server 1 as the transmission source address to the LAN 21. Alternatively, it is transmitted to the LAN interface 11.
[0059]
Thereby, the LAN switch 30 learns to transmit a packet addressed to the physical address of the LAN interface 2 to the port to which the LAN interface 11 is connected.
[0060]
As described above, according to the present embodiment, in addition to the effects of the first embodiment, there are the following effects. In other words, the packet filtering performed by the Ethernet switch reduces the number of packets arriving at the LAN interface of the standby server, and the processing of the standby server at the time of failure is less than in the first embodiment.
(Third embodiment)
As described in the prior art, when a logical address is taken over, the logical address can be taken over without affecting the client 22 on the LAN.
[0061]
This embodiment will be described with reference to FIGS.
This embodiment has a system configuration in which a packet filter 40 is added to the configuration of the first embodiment described above. In the figure, components similar to those in the first embodiment are denoted by the same reference numerals as those in FIG. 1 and description thereof is omitted.
[0062]
First, the packet filter 40 will be described with reference to FIG.
As shown in FIG. 8, the packet filter 40 includes a plurality of ports 41 to which the LAN interface of the computer is connected, an address conversion table 42 for storing addresses of packets to be converted, and a packet arriving from each port, and monitoring addresses Is internally provided with a filter 43 that performs the rewriting based on the address conversion table 42.
[0063]
The packet filter 40 transmits a packet arriving from a port to all other ports in the same manner as a normal LAN 21 hub. At that time, when a packet having the address registered in the original address 51 of the address conversion table 42 arrives, the destination is replaced with the one recorded in the new address 52 and transmitted to all ports. To do. Also, the packet source address is converted in the same manner.
[0064]
In order to instruct the packet filter 40 to replace the address as described above, the packet filter 40 has a physical address that identifies itself.
In order for the computer connected to the LAN 21 to control the packet filter 40, the original address and the new address are stored in the packet and transmitted to the physical address of the packet filter 40.
[0065]
The packet filter 40 receives the packet and registers it in the address conversion table 42. When a packet including data whose new address is the same address as the original address (address value is A) is received, the entry in the address conversion table 42 is searched for the original address A. Delete it.
[0066]
The packet filter 40 may be configured to have the LAN switch function described in the second embodiment.
The communication managers 4 and 13 operate in the same manner as in the first embodiment during normal times.
[0067]
Next, the time of failure will be described. As in the first embodiment, it is assumed that the primary server 1 has failed. The failure notification, the state change of the communication management units 4 and 13, and the operation of the communication management unit 4 of the primary server 1 are the same as in the first embodiment.
[0068]
The difference is the operation and position in place of communication performed by the communication management unit 13 of the standby server 10. This will be described with reference to FIG.
The communication managers 4 and 13 are located between the transport and the server process.
[0069]
When the communication management unit 13 receives the failure notification from the communication management unit 4, the communication management unit 13 instructs the transport 14 to receive a packet addressed to the logical address (IP address in TCP / IP) of the transport 5 of the primary server 1. To do.
[0070]
Since most of the transports 14 such as TCP / IP that are generally used can handle a plurality of logical addresses, a special mechanism is not necessary for this.
[0071]
In normal transport, in order to increase efficiency, other clients on the LAN cache a set of logical and physical addresses of machines on the LAN.
[0072]
When the standby server 10 starts using a new logical address, particularly in this embodiment, the cache address must be discarded in order to take over the logical address already paired with the physical address of the primary server 1. . In order to discard these caches, operations such as broadcasting and transmitting packets on the LAN are usually performed.
[0073]
There is no need to perform an operation for updating a cache of a set of logical addresses and physical addresses that a client on the LAN will hold. The reason will be described next.
[0074]
Next, the communication management unit 13 sets a set of data for the packet filter 40 with the physical address of the LAN interface 2 of the primary server 1 as the original address and the physical address of the LAN interface 11 of the standby server 10 as the new address. Send a packet containing
[0075]
The packet filter 40 receives this packet and records the set of physical addresses in the address conversion table 42. By doing so, the packet addressed to the LAN interface 2 is guaranteed to reach the LAN interface 11 by the packet filter 40. Accordingly, the client on the LAN continues to send packets to the LAN interface 2 of the primary server 1, but actually receives the standby server 10.
[0076]
Packets received by the LAN interface 11 reach the communication management unit 13 through the LAN driver 12 and the transport 14. The communication management unit 13 can determine whether the packet is addressed to the primary server or the standby server by looking at the logical address of the packet destination, and distributes the packet to each.
[0077]
Next, a case where the failure of the primary server 1 is corrected will be described. In this case, when it is detected that the failure has been corrected as in the first embodiment, the communication management unit 4 notifies the communication management unit 13.
[0078]
The communication management unit 13 notifies the transport 14 to stop handling the logical address of the primary server 1 that has taken over.
[0079]
Further, the packet filter 40 is instructed to transmit the packet that is the physical address of the LAN interface 2 of the primary server 1 to the packet filter 40 with both the original address and the new address, and stop the packet address conversion.
[0080]
Thus, the packet transfer by the packet filter 40 is stopped, the primary server 1 can receive the packet, and the standby server 10 can return to a state where only the packet addressed to itself can be received.
[0081]
According to the embodiment described above, the standby server can receive the packet of the primary server without affecting the logical address and physical address caches held by the client.
[0082]
Also, as described in the first and second embodiments, when a LAN interface that can specify only one physical address is used, it is not necessary to use a mode for receiving all packets.
(Fourth embodiment)
In all the embodiments described above, the primary server 1 and the standby server 10 exchange data for substituting communication by the standby server 10 using the dedicated communication link 20 in the event of a failure.
[0083]
In this embodiment, the primary server 1 and the standby server 10 are connected to a plurality of LANs, and even if a failure occurs in a plurality of LAN interfaces and communication becomes impossible, at least one can communicate. If there is a LAN, the above-mentioned failure notification and data exchange can be performed using a LAN instead of a dedicated communication link. According to this embodiment, no dedicated communication link is required between the primary server and the standby server.
[0084]
Note that. Since the technique of the present invention can be realized as software, it can be stored and distributed in a computer-readable recording medium such as a floppy disk, an optical disk, and a semiconductor memory as a program that can be executed by a computer. It is. Then, the computer that has read the contents of the recording medium implements the above-described processing by controlling the execution of the read program.
[0085]
【The invention's effect】
As described above in detail, according to the present invention, in order to change the combination of the logical address (IP address) of a transport (such as TCP / IP) and the physical address of the LAN, a notification is sent to other computers on the LAN. There is no need to put it out. Further, when a plurality of transports are used, processing is required for each transport. However, in the present invention, since transmission / reception of packets is substituted at the physical address level of the LAN, this is not necessary.
[0086]
Further, according to the present invention, by using the LAN switch, the number of packets that arrive at the standby LAN interface is reduced, and the load on the standby server and the LAN interface can be further reduced.
[0087]
Further, according to the present invention, the physical address of the packet can be converted and can be dynamically performed.
Furthermore, according to the present invention, the logical address managed by the transport is taken over, but there is an effect that it is not necessary to change the combination of the logical address and the physical address for other clients on the LAN.
[Brief description of the drawings]
FIG. 1 is a block diagram showing the configuration of a system according to a first embodiment of the present invention.
FIG. 2 is a conceptual diagram showing a normal communication process in the embodiment.
FIG. 3 is a conceptual diagram showing communication processing when a failure occurs in the embodiment.
FIG. 4 is a state transition diagram in the embodiment.
FIG. 5 is a flowchart showing a processing procedure of a communication management unit in the embodiment.
FIG. 6 is a block diagram showing the configuration of a system using a LAN switch according to the second embodiment of the present invention.
FIG. 7 is a block diagram showing a configuration example using a packet filter according to a third embodiment of the present invention.
FIG. 8 is a diagram showing a configuration of a packet filter in the embodiment.
FIG. 9 is a diagram showing a configuration of an address conversion table in the embodiment.
FIG. 10 is a conceptual diagram showing communication processing when a failure occurs in the embodiment.
FIG. 11 is a conceptual diagram showing communication processing by a conventional system.
[Explanation of symbols]
1 ... Main server,
2 ... LAN interface
3 ... LAN driver,
4. Communication management department
5 ... Transport,
6 ... Cluster controller,
7 ... Server process,
10: Standby server,
11 ... LAN interface,
12 ... LAN driver,
13 ... Communication management department,
14 ... Transport,
16 ... Server process,
20 ... communication link,
21 ... LAN,
22 ... Client,
30 ... LAN switch,
40 ... Packet filter,
41 ... Port,
42 ... address conversion table,
43. Filter.

Claims (2)

In a client server system having a main server and a standby server,
A dedicated communication link connecting the primary server and the standby server;
It is provided between the main server and the standby server and the LAN, has a plurality of ports to which the LAN interface of the computer is connected, and stores the physical address of the LAN interface of each port destination. A LAN switch for transferring packets addressed to the stored physical address taken in from the LAN only to the LAN interface of the corresponding port destination among the plurality of ports;
Comprising
The primary server is
When a failure that makes it impossible to execute communication processing via the LAN using the local server's LAN interface occurs, the physical address of the local server's LAN interface is notified to the standby server via the dedicated communication link. And means for requesting the standby server via the dedicated communication link to perform transmission and reception of packets for communication processing via the LAN via the dedicated communication link,
The standby server is
A mode for receiving all packets from the LAN when the primary server is requested via the dedicated communication link to send and receive packets for communication processing via the LAN via the dedicated communication link. The LAN interface of the own server is set in the local server, and a packet having the physical address of the LAN interface of the primary server notified via the dedicated communication link as a transmission source address is transmitted using the LAN interface of the own server. And a means for causing the LAN switch to learn to forward a packet addressed to the physical address of the LAN interface of the primary server to the LAN interface of the local server.
A client-server system. Provided between a primary server, a standby server, a dedicated communication link connecting the primary server and the standby server, and between the primary server, the standby server, and the LAN A plurality of ports to which the LAN interface of the computer is connected, storing the physical address of the LAN interface of each port destination, and receiving the packet addressed to the stored physical address taken from the LAN A communication management method for a client server system having a LAN switch that transfers only to a LAN interface of a corresponding port among the ports,
The primary server is
When a failure that makes it impossible to execute communication processing via the LAN using the local server's LAN interface occurs, the physical address of the local server's LAN interface is notified to the standby server via the dedicated communication link. Requesting the standby server via the dedicated communication link to perform transmission and reception of packets for communication processing via the LAN via the dedicated communication link;
The standby server is
A mode for receiving all packets from the LAN when the primary server is requested via the dedicated communication link to send and receive packets for communication processing via the LAN via the dedicated communication link. The LAN interface of the own server is set in the local server, and a packet having the physical address of the LAN interface of the primary server notified via the dedicated communication link as a transmission source address is transmitted using the LAN interface of the own server. And the LAN switch learns the packet addressed to the physical address of the LAN interface of the primary server to be transferred to the LAN interface of the local server.
A communication management method for a client-server system.

Images