Local logging in a distributed database management computer system

• US 5,966,706 A
• Filed: 02/19/1997
• Issued: 10/12/1999
• Est. Priority Date: 02/19/1997
• Status: Expired due to Fees

• Current Assignee: AT&T Corporation (AT&T, Inc.), Lucent Technologies, Inc. (Nokia Corporation)
• Original Assignee: AT&T Corporation (AT&T, Inc.), Lucent Technologies, Inc. (Nokia Corporation)
• Inventors: Rastogi, Rajeev R., Panagos, Euthimios, Biliris, Alexandros, Jagadish, Hosagrahar Visvesvaraya
• Primary Examiner(s): Amsbury, Wayne
• Assistant Examiner(s): ALAM, SHAHID AL
• Application Number:
  US08/803,042
• Time in Patent Office: 965 Days
• Field of Search: 707/3, 707/10, 707/101, 707/4, 707/2, 707/100, 707/102, 707/103, 707/200
• US Class Current: 1/1
• CPC Class Codes: G06F 16/2358 Change logging, detection, and notification replication G06F16/27
• CPC Class Codes: Y10S 707/99931 Database or file accessing
• CPC Class Codes: Y10S 707/99932 Access augmentation or optimizing
• CPC Class Codes: Y10S 707/99938 Concurrency, e.g. lock management in shared database
• CPC Class Codes: Y10S 707/99943 Generating database or data structure, e.g. via user interface

Abstract

A distributed database management computer system includes a plurality of nodes and a plurality of database pages. When a first node in the computer system updates a first database page, the first node generates a log record. The first node determines whether it manages the first database page. If the first node determines that it manages the first database page, the first node writes the log record to a log storage local to the first node. However, if the first node determines that it does not manage the first database page, the first node then determines whether it includes a local log storage. If the first node includes a local log storage, the first node writes the log record to the local log storage, even if the first node does not manage the first database page. If the first node does not include a local log storage, the first node sends the log record to a second node managing the first database page.

Claims

1. A method of operating a first node in a distributed database management computer system having a first set of database pages managed by the first node and a second set of database pages managed by a second node, wherein said first node can update pages in the first and second set, comprising the steps of:

(a) updating a database page at the first node;

(b) generating a first log record at the first node;

(c) determining whether the first database page is managed by the first node;

(d) if at step (c) it is determined that the first database page is managed by the first node, writing the first log record to a log storage local to the first node; and

(e) if at step (c) it is determined that the first database page is not managed by the first node;

(e-1) determining whether the first node includes a local log storage;

(e-2) if at step (e-1) it is determined that the first node includes a local log storage, writing the first log record to the local log storage; and

(e-3) if at step (e-1) it is determined that the first node does not include a local log storage, sending the first log record to a node that manages the first database page.

2. The method of claim 1, wherein each of the plurality of nodes includes lock information, further comprising the steps of:

(f) recovering from a failure of the first node by performing the following steps;

(f-1) determining which of the plurality of database pages require recovery;

(f-2) identifying which of the plurality of nodes are involved in the recovery;

(f-3) reconstructing the lock information; and

(f-4) coordinating the recovery among the involved nodes.

3. The method of claim 1, wherein the first node comprises a cache memory for buffering the log record and wherein either step (d) or step (e) is performed when said cache memory is full.

4. The method of claim 3, wherein either step (d) or step (e) is performed when the first database page is written to storage local to the first node.

5. The method of claim 3, wherein step (e) is performed when the first database page is sent to a second node that manages the first database page.

6. The method of claim 1, wherein either step (d) or step (e) is performed at a transaction commit.

7. A first node in a distributed database management computer system having a first set of database pages managed by the first node and a second set of database pages managed by a second node, wherein said first node can update pages in the first and second set, comprising:

means for updating a first database page at the first node;

means for generating a first log record at the first node;

first means for determining whether the first database page is managed by the first node;

first means for writing the first log record to a log storage local to the first node if said first means for determining determines that the first database page is managed by the first node;

second means for determining whether the first node includes a local log storage if said first means for determining determines that the first database page is not managed by the first node;

second means for writing the first log record to the local log storage if said second means for determining determines that the first node includes a local log storage; and

means for sending the first log record to a a node that manages the first database page if said second means for determining determines that the first node does not include a local log storage.

8. The computer system of claim 7, wherein each of the plurality of nodes includes lock information, further comprising:

means for recovering from a failure of the first node, said means for recovering comprising;

means for determining which of the plurality of database pages require recovery;

means for identifying which of the plurality of nodes are involved in the recovery;

means for reconstructing the lock information; and

means for coordinating the recovery among the involved nodes.