Highly available TCP systems with fail over connections

• US 6,871,296 B2
• Filed: 12/29/2000
• Issued: 03/22/2005
• Est. Priority Date: 12/29/2000
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Kashyap, Vivek
• Primary Examiner(s): LE, DIEU MINH T
• Application Number:
  US09/752,841
• Publication Number:
  US 20020087912A1
• Time in Patent Office: 1,544 Days
• Field of Search: 714/13, 714/2, 714/4, 714/16, 714/43, 714/44, 714/56, 714/20, 714/39, 709/105, 709/221, 709/227, 709/228
• US Class Current: 714/13
• CPC Class Codes: G06F 11/2005 using redundant communication controllers
• CPC Class Codes: G06F 11/2007 using redundant communication media
• CPC Class Codes: H04L 12/1877 Measures taken prior to transmission
• CPC Class Codes: H04L 69/40 for recovering from a failure of a protocol instance or entity, e.g. service redundancy protocols, protocol state redundancy or protocol service redirection management of faults, events, alarms or notifications in data switching networks H04L41/06

Abstract

The present invention provides a method, system and apparatus by which TCP connections may be failed-over from one system to another within a highly available network service, and appear transparent to the remote client. The connection state and ownership information of a system is broadcast within the network, so that if a first system crashes while running an application, a predetermined take-over policy causes a peer system to assume connection without loss of data such that a permanent connection has been established from the client'"'"'s point of view. After the failed system has been restored to a normal state, new connections are established by the first system.

Claims

1. A method by which a second system maintains connections for a failed first system, the second system and the first system being peers to one another, the method performed by the second system and comprising:

receiving ownership information from the first system on which an application is running, the ownership information indicating that the second system is to assume a connection of the first system upon failure of the first system;

prior to the first system having entered a failed state, snooping the connection from a client to the first system to use the application thereof, in order to maintain information regarding the connection as known by the first system;

determining that the first system is in the failed state; and

in response to determining that the first system is in the failed state, assuming the connection for the first system in a peer-to-peer manner,

wherein the snooping of the connection from the client to the first system enables the second system to assume the connection without losing the information regarding the connection as known by the first system, such that the client is unaware that the second system has assumed the connection for the first system.

2. The method of claim 1, wherein the ownership information is received when the second system is booted.

3. The method of claim 1, wherein the ownership information comprises:

a fail over policy;

a set of one or more IP addresses owned by the system;

a range of port numbers;

an application that is currently running on the system;

a ARP ownership policy;

a current protocol;

a set of one or more MAC addresses tied to the IP address(es); and

a cluster node ID.

4. The method of claim 1, wherein the first system and the second system solicit ownership information from each other.

5. The method of claim 1, further comprising determining that the first system has returned from the failed state to a normal state;

and returning responsibility for new connections to the first system.

6. The method of claim 1, wherein the first system and the second system are within a cluster of systems.

7. The method of claim 1, wherein the connections are TCP connections.

8. The method of claim 1, further comprising snooping connection data of the first system.

9. The method of claim 1, further comprising continuing the application from the point at which the first system failed.

10. The method of claim 1, further comprising:

attempting by a third system to assume the connection; and

sending from the third system to the second system a request to assume the connection.

11. A peer-to-peer system for maintaining a connection within a network, comprising:

means for broadcasting ownership information between a first system on which an application is running to at least a second system within the network, the first system and the second system being peers to one another, the ownership information indicating that the second system is to assume a connection of the first system by a client to use the application thereof, upon failure of the first system;

means for determining that the second system will assume the connection for the first system if the first system fails;

means for snooping by the second system prior to failure of the first system, packets sent to and received by the first system, so that the second system is able to maintain information regarding the connection as known by the first system;

means for determining that the first system is in a failed state;

means for continuing the application on the second system from the point at which the first system failed, the second system assuming the connection in a peer-to-peer manner, the snooping of the connection from the client to the first system enabling the second system to assume the connection without losing the information regarding the connection as known by the first system, such that the client is unaware that the second system has assumed the connection for the first system.

12. The system of claim 11, further comprising a means for broadcasting ownership information when each system is booted.

13. The system of claim 11, wherein the ownership information comprises:

a fail over policy;

a set of IP addresses owned by each system within the network;

a range of port numbers;

an application that is currently running on each system;

a ARP ownership policy;

a current protocol;

a set of MAC addresses tied to the IP address(es); and

a cluster node ID.

14. The system of claim 11, further comprising:

means for returning the first system from a failed state to a normal state; and

means by which connection is regained by the first system from the second system.

15. An article for maintaining connections by a second system for a first system, the first system and the second system being peers to one another, comprising:

a computer-readable signal-bearing medium;

means in the medium for receiving ownership information of the first system by the second system, the ownership information indicating that the second system is to assume a connection of the first system by a client to use an application thereof, upon failure of the first system,

means in the medium for snooping the connection by the second system prior to failure of the first system so that the second system is able to maintain information regarding the connection as known by the first system;

means in the medium for determining that the first system is in a failed state; and

means in the medium for assuming the connection for the first system by the second system in a peer-to-peer manner, the snooping of the connection from the client to the first system enabling the second system to assume the connection without losing the information regarding the connection as known by the first system, such that the client is unaware that the second system has assumed the connection for the first system.

16. The article of claim 15, further including:

means in the medium for continuing an application on the second system from the point at which the first system failed.

17. The article of claim 15, wherein the ownership information of the first system is broadcast when the system is booted.

18. The article of claim 15, wherein the ownership information comprises:

a fail over policy;

a set of IP addresses owned by the first system;

a range of port numbers;

an application that is currently running on the first system;

a ARP ownership policy;

a current protocol;

a set of MAC addresses tied to the IP address(es); and

a cluster node ID.

19. The article of claim 15, further comprising:

means in the medium for determining that the first system has returned from a failed state to a normal state; and

means in the medium for returning responsibility for connections to the first system.