Fault tolerant recoverable TCP/IP connection router

• US 6,148,410 A
• Filed: 09/15/1997
• Issued: 11/14/2000
• Est. Priority Date: 09/15/1997
• Status: Expired due to Term

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Levy-Abegnoli, Eric Michel, Schmidt, Donald William, Goldszmidt, German Sergio, Baskey, Michael Edward, Hunt, Guerney Douglass Holloway, Nick, Jeffrey Mark, Dillenberger, Donna Ngar-Ting
• Primary Examiner(s): Wright, Norman M.
• Application Number:
  US08/929,409
• Time in Patent Office: 1,156 Days
• Field of Search: 714/3, 714/4, 714/47, 714/56, 714/43, 714/48
• US Class Current: 714/4.11
• CPC Class Codes: H04L 41/0654 using network fault recovery ring fault isolation or reconfiguration in loop networks without recovery actions by a network management system H04L12/437
• CPC Class Codes: H04L 43/0811 by checking connectivity
• CPC Class Codes: H04L 45/00 Routing or path finding of packets in data switching networks routing or path finding in wireless networks H04W40/00
• CPC Class Codes: H04L 45/28 using route fault recovery
• CPC Class Codes: H04L 45/58 Association of routers
• CPC Class Codes: H04L 49/557 Error correction, e.g. fault recovery or fault tolerance

Abstract

A fault tolerant recoverable connection device and methods are disclosed that include a primary router in an active state to provide a connection between clients and servers in a network, and a backup router in a standby state. The states of the primary and backup routers are switched when the primary router fails, the backup router has a better reach-ability state than the primary router, or by an operator command. Each router has a synchronization manager which maintains synchronized tables between the active and standby routers; a monitoring manager which monitors and switch the state of the routers; a reach-ability manager which monitors, updates, and compares the reach-ability set and state of the routers; and a keep alive manager which monitors the routers to provide a status notification to the monitoring manager.

Claims

1. A fault tolerant recoverable transport connection protocol/Internet protocol (TCP/IP) connection device to provide a connection between clients and one or more servers in a computer network, said TCP/IP connection device comprising at least two TCP/IP routers connected between said clients and said one or more servers, wherein when a first TCP/IP router of said at least two TCP/IP routers is in an active state to provide said connection between said clients and said one or more servers, then a second TCP/IP router of said at least two TCP/IP routers is in a standby state and capable of providing connection between said clients and said servers with same or different connectivity than said first TCP/IP router;

each of said at least two TCP/IP routers including;

a synchronization manager that synchronizes internal tables of said at least two TCP/IP routers;

a monitoring manager that monitors a state of said at least two TCP/IP routers;

upon detecting a failure of said first TCP/IP router, said monitoring manager switching said second TCP/IP router from said standby state to said active state to maintain said connection, and switching said first TCP/IP router from said active state to said standby state; and

,

a reach-ability manager which compares a TCP/IP router reach-ability state of each of said at least two TCP/IP routers with a required reach-ability set comprising a set of hosts or subnets to which connections are to be maintained, and which instructs said monitoring manager to switch states of said at least two TCP/IP routers when said second TCP/IP router is in the standby state and has a reach-ability state that includes said required reach-ability set and is better than said reach-ability set of said active TCP/IP router, whereby TCP/IP connectivity is maintained between said clients and said servers regardless of said connectivity configuration.

2. The fault tolerant recoverable TCP/IP connection device of claim 1, wherein each of said at least two TCP/IP routers further comprises a reach-ability manager which updates said router reach-ability state.

3. The fault tolerant recoverable TCP/IP connection device of claim 1, wherein each of said at least two TCP/IP routers further comprises a keep alive manager that monitors said at least two TCP/IP routers to provide a status notification to said monitoring manager.

4. The fault tolerant recoverable TCP/IP connection device of claim 1, wherein said monitoring manager is integrated into each of said at least two TCP/IP routers.

5. The fault tolerant recoverable TCP/IP connection device of claim 1, wherein said active and standby states are determined by workload information.

6. The fault tolerant recoverable TCP/IP connection device of claim 1, wherein said first router is a primary TCP/IP router and said second router is a backup TCP/IP router, said primary TCP/IP router and said backup TCP/IP router are determined by workload information.

7. The fault tolerant recoverable TCP/IP connection device of claim 1, wherein said at least two TCP/IP routers include a first primary TCP/IP router for a first cluster of said one or more servers, a second primary TCP/IP router for a second cluster of said one or more servers, and a backup TCP/IP router shared by said first and second cluster of said one or more servers.

8. The fault tolerant recoverable TCP/IP connection device of claim 1, wherein said first TCP/IP router is a primary TCP/IP router for a first cluster of said one or more servers and a backup router for a second cluster of said one or more servers, and said second TCP/IP router is a primary TCP/IP router for said second cluster of said one or more servers and a backup TCP/IP router for said first cluster of said one or more servers.

9. A fault tolerant recoverable transport connection protocol/Internet protocol (TCP/IP) connection device to provide a connection between clients and servers in a computer network, said TCP/IP connection device comprising a primary TCP/IP router and one or more backup TCP/IP routers which are connected between said clients and servers and capable of providing connection between said clients and said servers with same or different connectivity than said first TCP/IP router, said primary TCP/IP router and one of one or more said backup TCP/IP routers switching between an active state and a standby state to maintain either one of said primary TCP/IP router and one of said one or more backup TCP/IP routers in the active state for providing said connection between said clients and said servers;

each of said primary TCP/IP router and said backup TCP/IP routers including a synchronization manager, a reach-ability manager, and a monitoring manager;

wherein synchronization managers of said primary TCP/IP router and said one of said one or more backup TCP/IP router communicate with each other to synchronize internal tables of said primary TCP/IP router and said one of said one or more backup TCP/IP router;

said monitoring managers of said primary TCP/IP router and said one of said one or more backup TCP/IP routers communicate with each other to monitor the state of said primary TCP/IP router and said one of said one or more backup TCP/IP router and perform the switching;

said reach-ability managers of one of said primary TCP/IP and said one of said one or more back-up TCP/IP routers being active for collecting from said primary TCP/IP router and said one of said one or more backup TCP/IP router a reach-ability state, comparing the reach-ability states of each of said primary TCP/IP router and said one of said one or more backup TCP/IP router to determine which of said primary TCP/IP router and said one of said one or more backup TCP/IP router has optimum reach-ability to elements of a required reach-ability set; and

, informing the monitoring manager for performing the switching in response to a TCP/IP router having an optimum reach-ability, whereby TCP/IP connectivity is maintained between said clients and said servers regardless of said connectivity configuration.

10. The fault tolerant recoverable TCP/IP connection device of claim 9, wherein each of said backup TCP/IP routers in the standby state monitors activities of said primary TCP/IP router in the active state so that internal tables of said primary TCP/IP router and said backup TCP/IP router are synchronized.

11. The fault tolerant recoverable TCP/IP connection device of claim 9, wherein the monitoring manager monitors the state of said primary TCP/IP router and said backup TCP/IP router for detecting a failure of one of said primary TCP/IP router and said backup TCP/IP router which is in the active state.

12. The fault tolerant recoverable TCP/IP connection device of claim 9, wherein the monitoring manager performs the switching in response to an operator command.

13. The fault tolerant recoverable TCP/IP connection device of claim 9, wherein said reach-ability set comprises information including hosts and subnets reachable by said connection device, said reach-ability manager of a primary TCP/IP router executing a configurable function to decide which connection device has an optimum current reach-ability.

14. The fault tolerant recoverable connection device of claim 9, wherein each of said primary TCP/IP router and said one or more TCP/IP backup routers further comprises a keep alive manager that monitors said primary TCP/IP router and said one or more TCP/IP backup routers to provide a status notification to said monitoring manager.

15. The fault tolerant recoverable connection device of claim 9, wherein said monitoring manager is integrated into each of said primary TCP/IP router and said backup TCP/IP router.

16. The fault tolerant recoverable connection device of claim 9, wherein said primary TCP/IP router and said backup TCP/IP router are determined by workload information.

17. The fault tolerant recoverable connection device of claim 9, wherein said active and standby states are determined by workload information.

18. The fault tolerant recoverable connection device of claim 9 further comprising another primary TCP/IP router for a cluster of servers, wherein said backup TCP/IP router is shared by said servers and said cluster of servers.

19. The fault tolerant recoverable connection device of claim 9, wherein said primary TCP/IP router of said of servers is a backup TCP/IP router for a cluster of servers, and said backup TCP/IP router of said of servers is a primary TCP/IP router for said cluster of servers.

20. A method of recovering from a failed connection between clients and a cluster of one or more servers comprising the steps of:

synchronizing internal tables of at least two fault tolerant recoverable transport connection protocol/Internet protocol (TCP/IP) routers connected between said clients and servers, a predetermined list of clients and servers reachable by each TCP/IP router comprising a reach-ability set, wherein when a first TCP/IP router of said at least two routers is in an active state to interconnect said clients and servers, then a second TCP/IP router of said at least two TCP/IP routers is in a standby state and capable of providing connection between said clients and said servers with same or different connectivity than said first TCP/IP router;

monitoring a state of said at least two TCP/IP routers;

detecting a failure of said first TCP/IP router;

switching said second TCP/IP router from said standby state to said active state to maintain interconnection between said clients and servers;

switching said first TCP/IP router from said active state to said standby state; and

,

collecting from said first TCP/IP router and said second TCP/IP router current reach-ability states;

enabling one of said first and second TCP/IP routers in an active state to compare reach-ability states of each of said first TCP/IP router and said second TCP/IP router with a required reach-ability set;

determining which of said first TCP/IP router and said second TCP/IP router has an optimum reach-ability; and

switching a TCP/IP router from a stand-by to an active state in accordance with results of the determining step, whereby TCP/IP connectivity is maintained between said clients and said servers regardless of said connectivity configuration.

21. The method of claim 20, wherein the synchronizing step is performed by a synchronization manager of each of said at least two TCP/IP routers, and the monitoring step, the detecting step and both switching steps are performed by a monitoring manager of said at least two TCP/IP routers.

22. The method of claim 21, wherein said step of switching a TCP/IP router from a stand-by to an active state further comprises the step of:

instructing said monitoring manager to switch states of said at least two TCP/IP routers when said second TCP/IP router is in the standby state and has a reach-ability state which is better than a reach-ability state of said first TCP/IP router.

23. The method of claim 22, wherein the comparing step and the instructing step are performed by a reach-ability manager of said at least two TCP/IP routers.

24. The method of claim 20 further comprising updating the internal tables with current reach-ability states.

25. The method of claim 24, wherein the updating step is performed by each reach-ability manager of said at least two TCP/IP routers.

26. The method of claim 20 further comprising the steps of:

monitoring said at least two TCP/IP routers; and

providing status notification to said monitoring manager.

27. The method of claim 26, wherein the monitoring step and the providing step are performed by a keep alive manager of said at least two TCP/IP routers.

28. A method of managing distribution of requests in a computer network of a type wherein a fault tolerant recoverable transport connection protocol/Internet protocol (TCP/IP) router distributes requests among multiple servers, comprising the steps of:

initializing a primary TCP/IP router in an active state and a backup TCP/IP router in a standby state, said backup TCP/IP router providing connection between said clients and said servers with same or different connectivity than said active TCP/IP router, said primary TCP/IP router and said backup TCP/IP router periodically receiving requests from the network and workload recommendations from the servers;

constructing a primary host connection table in the primary TCP/IP router and constructing a backup host connection table in the backup TCP/IP router based on the requests from the network and the workload recommendations, said workload recommendations being a function of capacities, availability and importance level of work running on the servers, the primary host connection table being synchronized with the backup host connection table;

distributing the requests according to the host connection table by the primary TCP/IP router in the active state;

sensing a failure from the primary TCP/IP router by a monitoring manager of one of the primary TCP/IP router and the backup TCP/IP router;

changing the backup TCP/IP router to the active state, in response to the failure;

distributing the requests by the backup TCP/IP router which is changed to the active state according to the backup host connection table in response to said changing the backup TCP/IP router to the active state; and

changing the primary TCP/IP router to the standby state by the monitoring manager of the backup TCP/IP router, whereby TCP/IP connectivity is maintained between said clients and said servers regardless of said connectivity configuration.

29. The method of claim 28, wherein the initializing step is performed by a monitoring manager of the primary TCP/IP router.

30. The method of claim 28, wherein the constructing step is performed by a synchronization manager of each the primary TCP/IP router and the backup TCP/IP router.

31. The method of claim 28 further comprising communicating configuration information between the primary and backup routers so that configuration information of the primary router need not be reconstructed by the backup router to prevent delays in recovery of requests distributions when the primary router fails.

32. A method of switching between active and standby states of fault tolerant recoverable transport connection protocol/Internet protocol (TCP/IP) routers in a network having a primary TCP/IP router in an active state for providing a connection between clients and servers and a backup TCP/IP router in a standby state and capable of providing connection between said clients and said servers with same or different connectivity than said primary TCP/IP router comprising the steps of:

providing a reach-ability manager at each said primary TCP/IP and back-up TCP/IP router;

computing reach-ability set and state of said active TCP/IP router and reach-ability set and state of said standby TCP/IP router, said reach-ability manager of said back-up TCP/IP router reporting a reach-ability state to said reach-ability manager of said active TCP/IP router;

said active TCP/IP router implementing a configurable function for comparing the reach-ability set and state of said active TCP/IP router with the reach-ability set and state of said standby TCP/IP router; and

switching said primary TCP/IP router from the active state to the standby state and said backup TCP/IP router from the standby state to the active state when the reach-ability set and state of said standby TCP/IP router are better than the reach-ability set and state of said active TCP/IP router, whereby TCP/IP connectivity is maintained between said clients and said servers regardless of said connectivity configuration.