Lease based safety protocol for distributed system with multiple networks

• US 6,775,703 B1
• Filed: 05/01/2000
• Issued: 08/10/2004
• Est. Priority Date: 05/01/2000
• Status: Expired due to Term

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Goel, Atul, Rees, Robert Michael, Long, Darrell D. E., Burns, Randal Chilton
• Primary Examiner(s): KANG, PAUL H
• Application Number:
  US09/562,510
• Time in Patent Office: 1,562 Days
• Field of Search: 709/220, 709/221, 709/222, 709/223, 709/225, 709/226, 709/228, 709/232, 709/238, 709/210, 709/227, 709/224, 713/200, 711/148, 711/147, 711/150, 711/153, 711/154
• US Class Current: 709/228
• CPC Class Codes: G06F 12/0815 Cache consistency protocols
• CPC Class Codes: G06F 12/0866 for peripheral storage systems, e.g. disk cache
• CPC Class Codes: H04L 67/1097 for distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS]
• CPC Class Codes: H04L 69/28 Timers or timing mechanisms used in protocols
• CPC Class Codes: H04L 69/329 in the application layer [OSI layer 7]
• CPC Class Codes: H04L 9/40 Network security protocols

Abstract

A system, method, and computer program product for a lease-based timeout scheme that addresses fencing'"'"'s shortcomings. Unlike fencing, this scheme (or protocol) enables an isolated computer to realize it is disconnected from the distributed system and write its dirty data out to storage before its locks are stolen. In accordance with the invention, data consistency during a partition in a distributed system is ensured by establishing a lease based protocol between in the distributed system wherein a client can hold a lease with a server. The lease represents a contract between a client and a server in which the server promises to respect the client for a period of time. The server respects the contract even when it detects a partition between the client and itself.

Claims

1. A computer program product comprising a machine-readable medium including machinxecutable instructions therein for ensuring data consistency in a distributed system in the presence of failures, comprising the steps of:

establishing a lease based protocol in said distributed system wherein a client may hold a lease with a server, wherein said lease represents a contract between said client and said server in which said server promises to respect said client for a period of time and wherein said server respects said contract even when said server detects a partition between itself and said client; and

renewing said lease in response to a client initiated message that has been acknowledged by said server, said client initiated message being other than a request for a lease renewal, wherein said lease based protocol executed by said client includes four phases;

a first phase in which said lease protects access to all data objects locked on said server and during which any new message ACK received renews said lease;

a second Phase, entered into in response to said client not receiving a message acknowledgement in said first phase, during which said client attempts to obtain a new lease by sending said server a keep-alive protocol message and during which said client continues to service file system requests on behalf of local processes;

a third phase, responsive to failure to attain said lease in said second phase, wherein said client stops servicing file system requests on behalf of said local processes but continues in-process file system transactions; and

a fourth phase, responsive to a failure to recover said lease during said third phase, wherein any dirty pages protected by locks held with said server for which said lease expires are flushed out to storage.

2. A method for ensuring data consistency in a distributed system in the presence of failures, comprising the steps of:

establishing a lease based protocol in said distributed system wherein a client may hold a lease with a server, wherein said lease represents a contract between said client and said server in which said server promises to respect said client for a period of time and wherein said server respects said contract even when said server detects a partition between itself and said client; and

renewing said lease in response to a client initiated message that has been acknowledged by said server, said client initiated message being other than a request for a lease renewals wherein said lease based protocol executed by said client includes four phases;

a first phase in which said lease protects access to all data objects locked on said server and during which any new message ACK received renews said lease;

a second phase, entered into in response to said client not receiving a message acknowledgement in said first phase, during which said client attempts to obtain a new lease by sending said server a keep-alive protocol message and during which said client continues to service file system requests on behalf of local processes;

a third phase, responsive to failure to attain said lease in said second phase, wherein said client stops servicing file system requests on behalf of said local processes but continues in-process file system transactions; and

a fourth phase, responsive to a failure to recover said lease during said third phase, wherein any dirty pages protected by locks held with said server for which said lease expires are flushed out to storage.

3. The method of claim 2, wherein said server maintains no state information concerning client failures except in response to detection of a partition.