RMI-IIOP request failover mechanism

• US 7,444,536 B1
• Filed: 08/16/2004
• Issued: 10/28/2008
• Est. Priority Date: 04/16/2004
• Status: Active Grant

• Current Assignee: Oracle America, Inc. (Oracle Corporation)
• Original Assignee: Sun Microsystems Incorporated (Oracle Corporation)
• Inventors: Jairath, Pankaj
• Primary Examiner(s): Beausoliel; Robert
• Assistant Examiner(s): Riad; Amine
• Application Number:
  US10/919,049
• Time in Patent Office: 1,534 Days
• Field of Search: 714/4
• US Class Current: 714/4.11
• CPC Class Codes: G06F 11/1482 by means of middleware or OS functionality
• CPC Class Codes: G06F 11/2046 where the redundant components share persistent storage G06F11/2043 takes precedence
• CPC Class Codes: G06F 9/465 Distributed object oriented systems remote method invocation [RMI] G06F9/548
• CPC Class Codes: H04L 67/1095 Replication or mirroring of data, e.g. scheduling or transport for data synchronisation between network nodes
• 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

Embodiments of a request failover mechanism may be used to failover requests from clients of server instances in a cluster to alternative server instances. One embodiment may use cluster-aware remote references that include the IIOP endpoints at which the application server instances forming the cluster listen to IIOP requests. Using a cluster-aware remote reference to an object on an application server instance, a request to the object may be failed-over to an alternate endpoint in the cluster when the primary endpoint is unreachable. In one embodiment, once a request for a referenced object is failed over to a new server instance, all the subsequent requests to that object will continue to go to the same server instance even if the original server instance becomes available. One embodiment may provide client-side IIOP request load balancing in clustered application server environments through a load balancer subsystem using the cluster-aware remote references.

Claims

1. An application server cluster system, comprising:

two or more application server instances in a cluster; and

an IIOP or RMI-IIOP client of one of the application server instances, wherein the client is configured to;

detect failure of the application server instance; and

failover to an alternate one of the application server instances in the cluster in response to said failure of the application server instance.

2. The system as recited in claim 1, wherein the client is configured to remain localized to the alternate application server instance after said failover for the lifetime of the client.

3. The system as recited in claim 1, wherein the client includes a remote reference indicating an endpoint on the application server instance, and wherein, to detect failure of the application server instance, the client is further configured to detect failure of a request to the endpoint on the application server instance.

4. The system as recited in claim 3, wherein, to failover to an alternate one of the application server instances, the client is further configured to determine an alternate endpoint on the alternate application server instance from the remote reference.

5. The system as recited in claim 1, wherein the client comprises a cluster-aware remote reference to an endpoint for an object on the application server instance, and wherein, to failover to an alternate one of the application server instances, the client is further configured to determine an alternate endpoint for a replica of the object on the alternate application server instance from the cluster-aware remote reference.

6. The system as recited in claim 5, wherein, after said failover to the alternate application server instance, the alternate application server instance is configured to restore a state of the object from the failed application server instance to the alternate application server instance.

7. The system as recited in claim 6, wherein the state of the object is restored from a backend database for storing states of objects in the cluster.

8. The system as recited in claim 1, wherein the two or more application server instances are homogeneous.

9. The system as recited in claim 1, wherein the application server instance is configured to provide a cluster-aware remote reference to an object on the application server instance to the client.

10. The system as recited in claim 9, wherein the cluster-aware remote reference indicates a primary endpoint on the application server instance and one or more alternate endpoints on the other application server instances.

11. The system as recited in claim 9, wherein the object is one of an EJB or a Name Service.

12. The system as recited in claim 1, wherein the application server instance comprises a server configuration file comprising configuration information for all the application server instances in the cluster, wherein the configuration information includes endpoint information for all the application server instances in the cluster.

13. The system as recited in claim 12, wherein the application server instance is configured to provide a cluster-aware remote reference to an object on the application server instance to the client, wherein the cluster-aware remote reference indicates a primary endpoint on the application server instance and one or more alternate endpoints on the other application server instances obtained from the server configuration file.

14. The system as recited in claim 1, wherein the client is one of an Application Client Container (ACC) client and a standalone client.

15. The system as recited in claim 1, wherein the client is further configured to load balance requests to the application server instances in the cluster.

16. The system as recited in claim 1, wherein the client is further configured to load balance requests to the application server instances according to one of an initial context-based load balancing policy and a lookup-based load balancing policy.

17. The system as recited in claim 1, wherein the client comprises a cluster-aware remote reference to the application server instances, wherein the cluster-aware remote reference indicates endpoints on the application server instances in the cluster, and wherein the client is further configured to load balance requests to the application server instances in the cluster according to a load balancing policy that is configured to select an endpoint from among the endpoints indicated in the cluster-aware remote reference.

18. The system as recited in claim 1,

wherein the client is further configured to provide a failover mapping indicating one or more remote references associated with a failed-over cluster-aware remote reference to the alternate application server instance;

and

wherein the alternate application server instance is configured to access the failover mapping to dispatch one or more outgoing invocations to the other remote references indicated by a failed-over request to the failed-over cluster-aware remote reference.

19. An application server cluster system, comprising:

two or more application server instances in a cluster; and

an IIOP or RMI-IIOP client of one of the application server instances, wherein the client is configured to;

obtain a cluster-aware remote reference to an object on the application server instance from the application server instance, wherein the cluster-aware remote reference indicates a primary endpoint for accessing an instance of the object on the application server instance and one or more alternate endpoints for accessing other instances of the object on the other application server instances;

detect a communication failure for a request to the primary endpoint for accessing the instance of the object on the application server instance; and

failover the request to one of the alternate endpoints on one of the other application server instances in response to said communication failure.

20. The system as recited in claim 19, wherein, after said failover to the alternate application server instance, the cluster-aware remote reference remains localized to the alternate application server instance for the lifetime of the client.

21. The system as recited in claim 19, wherein requests on the cluster-aware remote reference continue to go only to the alternate endpoint on the other application server instance if the failed application server instance subsequently becomes available in the cluster.

22. The system as recited in claim 19, wherein, after said failover to the alternate application server instance, the other application server instance is configured to restore a state of the object from the failed application server instance to the other application server instance.

23. The system as recited in claim 22, wherein the state of the object is restored from a backend database for storing states of objects in the cluster.

24. The system as recited in claim 19, wherein the two or more application server instances are homogeneous.

25. The system as recited in claim 19, wherein the object is one of an EJB or a Name Service.

26. The system as recited in claim 19, wherein the application server instance comprises a server configuration file comprising configuration information for all the application server instances in the cluster, wherein the configuration information includes endpoint information for all the application server instances in the cluster.

27. The system as recited in claim 26, wherein the primary endpoint on the application server instance and the one or more alternate endpoints on the other application server instances are obtained from the server configuration file.

28. The system as recited in claim 19, wherein the client is one of an Application Client Container (ACC) client and a standalone client.

29. The system as recited in claim 19, wherein the client is further configured to load balance requests to the application server instances in the cluster.

30. The system as recited in claim 19, wherein the client is further configured to load balance requests to the application server instances in the cluster according to a load balancing policy that is configured to select an endpoint from among the endpoints indicated in the cluster-aware remote reference.

31. The system as recited in claim 19, wherein the client is configured to load balance requests to the application server instances according to one of an initial context-based load balancing policy and a lookup-based load balancing policy.

32. The system as recited in claim 19,

wherein the client is further configured to provide a failover mapping indicating one or more other remote references associated with the cluster-aware remote reference to the alternate application server instance during said failover;

and

wherein the alternate application server instance is configured to access the failover mapping to dispatch one or more outgoing invocations to the other remote references indicated by the failed-over request.

33. A system, comprising:

a processor; and

a memory comprising program instructions, wherein the program instructions are executable by the processor to implement an IIOP or an RMI-IIOP client application for accessing a cluster comprising a plurality of application server instances, wherein the client application is configured to;

obtain a cluster-aware remote reference to an object on the application server instances from the cluster, wherein the cluster-aware remote reference indicates a primary endpoint for accessing an instance of the object on one of the application server instances and one or more alternate endpoints for accessing other instances of the object on other application server instances in the cluster;

detect a communication failure for a request to the primary endpoint for accessing the instance of the object on the application server instance; and

failover the request to one of the alternate endpoints on one of the other application server instances in response to said communication failure.

34. The system as recited in claim 33, wherein, after said failover to the alternate application server instance, the cluster-aware remote reference remains localized to the alternate application server instance for the lifetime of the client.

35. The system as recited in claim 33, wherein the object is one of an EJB or a Name Service.

36. The system as recited in claim 33, wherein the client is one of an Application Client Container (ACC) client and a standalone client.

37. The system as recited in claim 33, wherein the client is further configured to load balance requests to the application server instances in the cluster according to a load balancing policy that is configured to select an endpoint from among the endpoints indicated in the cluster-aware remote reference.

38. The system as recited in claim 33, wherein the client is configured to load balance requests to the application server instances according to one of an initial context-based load balancing policy and a lookup-based load balancing policy.

39. A system, comprising:

a processor; and

a memory comprising program instructions, wherein the program instructions are executable by the processor to implement an application server instance in a cluster of application server instances, wherein the application server instance is configured to;

receive a failed-over request to an object on a failed application server instance in the cluster from a client of the failed application server instance; and

restore a state of the object on the failed application server instance to the application server instance in response to said failed-over request received from the client.

40. The system as recited in claim 39, wherein the state of the object is restored from a backend database for storing states of objects on the application server instances in the cluster.

41. The system as recited in claim 39, wherein the application server instances in the cluster are homogeneous.

42. The system as recited in claim 39, wherein the object is one of an EJB or a Name Service.

43. The system as recited in claim 39, wherein the application server instance is further configured to generate cluster-aware remote references for IIOP and RMI-IIOP clients of the application server instance.

44. The system as recited in claim 39, wherein the application server instance is configured to maintain a server configuration file comprising configuration information for all the application server instances in the cluster, wherein the configuration information includes endpoint information for accessing objects on all the application server instances in the cluster, wherein the application server instance is further configured to generate cluster-aware remote references for IIOP and RMI-IIOP clients of the application server instance from the server configuration file.

45. The system as recited in claim 39, wherein the client is one of an Application Client Container (ACC) client and a standalone client.

46. A system, comprising:

client-side means for failing over remote references for IIOP and RMI-IIOP clients of a cluster of application server instances from a failed one of the application server instances to an alternate application server instance; and

means for load balancing requests from the IIOP and RMI-IIOP clients among the application server instances in the cluster.

47. A method, comprising:

providing a cluster-aware remote reference to an object on an application server instance in a cluster of two or more application server instances to an IIOP or RMI-IIOP client of the application server instance, wherein the cluster-aware remote reference indicates a primary endpoint on the application server instance and one or more alternate endpoints on other application server instances in the cluster;

the client issuing a request to the object via the cluster-aware remote reference;

detecting a communication failure for the request, wherein the communication failure indicates failure of the application server instance;

selecting an alternate endpoint on one of the other application server instances from the one or more alternate endpoints indicated by the cluster-aware remote reference; and

failing over the cluster-aware remote reference to the alternate endpoint on the other application server instance.

48. The method as recited in claim 47, wherein the cluster-aware remote reference remains localized to the other application server instance for the lifetime of the client.

49. The method as recited in claim 47, further comprising:

the failed application server instance becoming available in the cluster;

a subsequent request on the cluster-aware remote reference going only to the alternate endpoint on the other application server instance after the failed application server instance becoming available in the cluster.

50. The method as recited in claim 47, further comprising restoring a state of the object from the failed application server instance to the other application server instance.

51. The method as recited in claim 50, wherein the state of the object is restored from a backend database for storing states of objects in the cluster.

52. The method as recited in claim 47, wherein the two or more application server instances are homogeneous.

53. The method as recited in claim 47, wherein the object is one of an EJB or a Name Service.

54. The method as recited in claim 47, wherein the application server instance comprises a server configuration file comprising configuration information for all the application server instances in the cluster, wherein the configuration information includes endpoint information for all the application server instances in the cluster.

55. The method as recited in claim 54, further comprising obtaining the primary endpoint on the application server instance and the one or more alternate endpoints on the other application server instances from the server configuration file.

56. The method as recited in claim 47, wherein the client is one of an Application Client Container (ACC) client and a standalone client.

57. The method as recited in claim 47, further comprising the client load balancing requests to the application server instances in the cluster.

58. The method as recited in claim 47, further comprising the client load balancing requests to the application server instances in the cluster according to a load balancing policy for selecting an endpoint from among the endpoints indicated in the cluster-aware remote reference.

59. The method as recited in claim 47, further comprising the client load balancing requests to the application server instances according to one of an initial context-based load balancing policy and a lookup-based load balancing policy.

60. The method as recited in claim 47, further comprising:

the client providing a failover mapping indicating one or more other remote references associated with the failed-over cluster-aware remote reference to the other application server instance; and

the other application server instance accessing the failover mapping to dispatch one or more outgoing invocations to the other remote references indicated by the failed-over request.

61. A computer-readable storage medium, storing program instructions, wherein the program instructions are computer-executable to implement:

providing a cluster-aware remote reference to an object on an application server instance in a cluster of two or more application server instances to an IIOP or RMI-IIOP client of the application server instance, wherein the cluster-aware remote reference indicates a primary endpoint on the application server instance and one or more alternate endpoints on other application server instances in the cluster;

the client issuing a request to the object via the cluster-aware remote reference;

detecting a communication failure for the request, wherein the communication failure indicates failure of the application server instance;

selecting an alternate endpoint on one of the other application server instances from the one or more alternate endpoints indicated by the cluster-aware remote reference; and

failing over the cluster-aware remote reference to the alternate endpoint on the other application server instance.

62. The computer-readable storage medium as recited in claim 61, wherein the cluster-aware remote reference remains localized to the other application server instance for the lifetime of the client.

63. The computer-readable storage medium as recited in claim 61, wherein the program instructions are further computer-executable to implement:

the failed application server instance becoming available in the cluster;

a subsequent request on the cluster-aware remote reference going only to the alternate endpoint on the other application server instance after the failed application server instance becoming available in the cluster.

64. The computer-readable storage medium as recited in claim 61, wherein the program instructions are further computer-executable to implement restoring a state of the object from the failed application server instance to the other application server instance.

65. The computer-readable storage medium as recited in claim 64, wherein the state of the object is restored from a backend database for storing states of objects in the cluster.

66. The computer-readable storage medium as recited in claim 61, wherein the two or more application server instances are homogeneous.

67. The computer-readable storage medium as recited in claim 61, wherein the object is one of an EJB or a Name Service.

68. The computer-readable storage medium as recited in claim 61, wherein the application server instance comprises a server configuration file comprising configuration information for all the application server instances in the cluster, wherein the configuration information includes endpoint information for all the application server instances in the cluster.

69. The computer-readable storage medium as recited in claim 68, wherein the program instructions are further computer-executable to implement obtaining the primary endpoint on the application server instance and the one or more alternate endpoints on the other application server instances from the server configuration file.

70. The computer-readable storage medium as recited in claim 61, wherein the client is one of an Application Client Container (ACC) client and a standalone client.

71. The computer-readable storage medium as recited in claim 61, wherein the program instructions are further computer-executable to implement the client load balancing requests to the application server instances in the cluster.

72. The computer-readable storage medium as recited in claim 61, wherein the program instructions are further computer-executable to implement the client load balancing requests to the application server instances in the cluster according to a load balancing policy for selecting an endpoint from among the endpoints indicated in the cluster-aware remote reference.

73. The computer-readable storage medium as recited in claim 61, wherein the program instructions are further computer-executable to implement the client load balancing requests to the application server instances according to one of an initial context-based load balancing policy and a lookup-based load balancing policy.

74. The computer-readable storage medium as recited in claim 61, wherein the program instructions are further computer-executable to implement:

the client providing a failover mapping indicating one or more other remote references associated with the failed over cluster-aware remote reference to the other application server instance; and

the other application server instance accessing the failover mapping to dispatch one or more outgoing invocations to the other remote references indicated by the failed-over request.