Selecting buckets for redistributing data between nodes in a parallel database in the quiescent mode

• US 5,634,125 A
• Filed: 09/02/1993
• Issued: 05/27/1997
• Est. Priority Date: 09/02/1993
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Li, Shih-Gong
• Primary Examiner(s): Black, Thomas G.
• Assistant Examiner(s): Von Buhr, Maria N.
• Application Number:
  US08/116,086
• Time in Patent Office: 1,363 Days
• Field of Search: 395/600, 395/425, 395/200, 395/650, 395/619, 364/DIG. 1, 364/DIG. 2
• US Class Current: 1/1
• CPC Class Codes: G06F 16/27 Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor
• CPC Class Codes: G06F 16/275 Synchronous replication
• CPC Class Codes: Y10S 707/99954 Version management

Abstract

A method, system and program for determining which data stored in the nodes in a parallel database to redistribute to a receiving node. First, an average workload for the nodes within the parallel database is calculated based on the total workload divided by the number of existing and new nodes. Next, a set of transmitting nodes is established each of which has a workload which exceeds the average workload. A set of data structures which when redistributed from the transmitting node to the receiving node will lower the transmitting node workload to equal or slightly greater than the average workload is selected. Finally, the set of data structures is redistributed from the transmitting node to the receiving node.

Claims

1. A method for determining which data stored in the nodes in a parallel database system to redistribute to a receiving node, comprising the steps of:

calculating an average workload for the nodes within the parallel database system;

determining a transmitting node having a workload which exceeds the average workload;

selecting redistributable data which when redistributed from the transmitting node to the receiving node will lower the transmitting node workload toward the average workload; and

redistributing the redistributable data from the transmitting node to the receiving node.

2. The method as recited in claim 1 wherein the selecting step further comprises the steps of:

ordering a list of redistributable candidate data sets according to a redistribution value of each respective candidate data set;

subtracting a workload associated with a first candidate data set having a highest redistribution value from the transmitting node workload; and

selecting the first candidate data set, as a redistributable data set, if a result of the subtracting step is greater than or equal to the average workload.

3. The method as recited in claim 2 wherein the steps are repeated for a plurality of transmitting nodes each of which has a workload greater than the average workload.

4. The method as recited in claim 1 wherein the receiving node is a new node to the parallel database system.

5. The method as recited in claim 1 wherein the receiving node is an underutilized node in the parallel database system.

6. A system for determining which data stored in the nodes in a parallel database system to redistribute to a receiving node, comprising:

means for calculating an average workload for the nodes within the parallel database;

means for determining a transmitting node having a workload which exceeds the average workload;

means for selecting redistributable data which when redistributed from the transmitting node to the receiving node will lower the transmitting node workload toward the average workload; and

means for redistributing the redistributable data from the transmitting node to the receiving node.

7. The system as recited in claim 6 wherein the selecting means further comprises:

means for ordering a list of redistributable candidate data sets according to a redistribution value of each respective candidate data set;

means for subtracting a workload associated with a first candidate data set having a highest redistribution value from the transmitting node workload; and

means for selecting the first candidate data set as a first redistributable data set, responsive to a determination that the transmitting node workload after redistributing the first candidate set would be greater than or equal to the average workload.

8. The system as recited in claim 7 wherein data sets from a plurality of transmitting nodes each of which has a workload greater than the average workload are redistributed.

9. The system as recited in claim 6 wherein the receiving node is a new node to the parallel database system.

10. The system as recited in claim 6 wherein the receiving node is an underutilized node in the parallel database system.

11. A computer program product in a computer readable memory for determining which data stored in the nodes in a parallel database system to redistribute to a receiving node, comprising:

means for calculating an average workload for the nodes within the parallel database;

means for determining a transmitting node having a workload which exceeds the average workload;

means for selecting redistributable data which when redistributed from the transmitting node to the receiving node will lower the transmitting node workload toward the average workload; and

means for redistributing the redistributable data from the transmitting node to the receiving node.

12. The product as recited in claim 11 wherein the selecting means further comprises:

means for ordering a list of redistributable candidate data sets according to a redistribution value of each respective candidate data set;

means for subtracting a workload associated with a first candidate data set having a highest redistribution value from the transmitting node workload; and

means for selecting the first candidate data set as a first redistributable data set, responsive to a determination that the transmitting node workload after redistributing the first candidate set would be greater than or equal to the average workload.

13. The product as recited in claim 12 wherein data sets from a plurality of transmitting nodes each of which has a workload greater than the average workload are redistributed.

14. The product as recited in claim 11 wherein the receiving node is a new node to the parallel database system.

15. The product as recited in claim 11 wherein the receiving node is an underutilized node in the parallel database system.