SYSTEM USING VIRTUAL REPLICATED TABLES IN A CLUSTER DATABASE MANAGEMENT SYSTEM

• US 20090043863A1
• Filed: 10/17/2008
• Published: 02/12/2009
• Est. Priority Date: 07/08/2004
• Status: Active Grant

• Current Assignee: Domo, Inc.
• Original Assignee: International Business Machines Corporation
• Inventors: Lashley, Scott David, Taylor, Paul Sherwood
• Granted Patent: US 8,032,488 B2
• Time in Patent Office: Days
• US Class Current: 709/213
• CPC Class Codes: G06F 12/0813 with a network or matrix configuration

Abstract

A system for improved data sharing within a cluster of nodes having a database management system. The system defines a virtual replicated table as being useable in a hybrid of a shared-cache and shared-nothing architecture. The virtual replicated table is a physically single table sharable among a plurality of cluster nodes for data read operations and not sharable with other cluster nodes for data modification operations. Default owner node is assigned for each virtual replicated table to ensure the page validity and provide requested pages to the requesting node.

Claims

1. A system for improved data sharing within a cluster of nodes having a database management system, comprising:

a network;

a node computer system connected to the network, the node computer system comprising;

a processor;

a memory coupled to the processor;

a database connected to the node computer system, the database comprising;

a virtual replicated table useable in a hybrid of a shared-cache and shared-nothing architecture and comprising a physically single table sharable among a plurality of cluster nodes for data read operations and the physically single table being not sharable with other cluster nodes for data modification operations;

means for performing a read operation on virtual replicated table data in a shared-cache mode in a local node using a local buffer cache, the read operation for each requested page performing lock synchronization, caching pages from the virtual replicated table into the local buffer cache, retrieving data from the page, and performing cleanup and page unlocking; and

means for performing a modification operation on the virtual replicated table data exclusively in a shared-nothing mode, the modification operation shipped to and executed exclusively in the owner node using a local buffer cache in a shared-nothing algorithm and performing lock synchronization, caching pages from the virtual replicated table into the local buffer cache, retrieving and modifying data from the page, and performing cleanup and page unlocking.

2. The system of claim 1, further comprising means for lock synchronization ensuring that records of the virtual replicated table being locked for read operations with a local transaction lock (TX-Lock) and for modification operations being locked with both local and global TX-Locks across the cluster nodes and wherein the TX-Locks being releasable upon operation completion.

3. The system of claim 1, further comprising means for locking index and internal system pages across cluster nodes using a system-level page-cache lock (P-Lock), the system-level page-cache lock releasable prior to operation completion when the protected page consistency is guaranteed across the cluster.

4. The system of claim 1, further comprising means for assigning a default owner node for each virtual replicated table wherein the owner node ensures the page validity and provides requested pages to a requesting node.

5. The system of claim 1, further comprising means for defining one or more partitioned tables in a cluster, the cluster comprising one or more virtual replicated tables, operations against the partitioned tables comprising standard shared-nothing architecture operations.

6. The system of claim 1, further comprising means for defining one or more non-partitioned tables in a cluster, the cluster comprising one or more virtual replicated tables, operations against the non-partitioned tables comprising standard shared-nothing architecture operations.

7. A computer program product comprising a computer storage medium tangibly embodying a program of instructions executable to perform operations for improved data sharing within a cluster of nodes having a database management system, the operations of the computer program product comprising:

defining a virtual replicated table in the database management system, the virtual replicated table useable in a hybrid of a shared-cache and shared-nothing architecture and comprising a physically single table sharable among a plurality of cluster nodes for data read operations and the physically single table being not sharable with other cluster nodes for data modification operations;

performing a read operation on virtual replicated table data in a shared-cache mode in a local node using a local buffer cache, the read operation for each requested page performing lock synchronization, caching pages from the virtual replicated table into the local buffer cache, retrieving data from the page, and performing cleanup and page unlocking; and

performing a modification operation on the virtual replicated table data exclusively in a shared-nothing mode, the modification operation shipped to and executed exclusively in the owner node using a local buffer cache in a shared-nothing algorithm and performing lock synchronization, caching pages from the virtual replicated table into the local buffer cache, retrieving and modifying data from the page, and performing cleanup and page unlocking.

8. The computer program product of claim 7, further comprising ensuring that records of the virtual replicated table are locked for read operations with a local transaction lock (TX-Lock) and modification operations are locked with both local and global TX-Locks across the cluster nodes and wherein the TX-Locks are releasable upon operation completion.

9. The computer program product of claim 7, further comprising locking index and internal system pages across cluster nodes using a system-level page-cache lock (P-Lock), the system-level page-cache lock releasable prior to operation completion when the protected page consistency is guaranteed across the cluster.

10. The computer program product of claim 7, further comprising assigning a default owner node for each virtual replicated table wherein the owner node ensures the page validity and provides requested pages to a requesting node.

11. The computer program product of claim 7, further comprising defining one or more partitioned tables in a cluster, the cluster comprising one or more virtual replicated tables, operations against the partitioned tables comprising standard shared-nothing architecture operations.

12. The computer program product of claim 7, further comprising defining one or more non-partitioned tables in a cluster, the cluster comprising one or more virtual replicated tables, operations against the non-partitioned tables comprising standard shared-nothing architecture operations.