Incremental run-time session balancing in a multi-node system

• US 20050055446A1
• Filed: 08/12/2004
• Published: 03/10/2005
• Est. Priority Date: 08/14/2003
• Status: Active Grant

• Current Assignee: Oracle International Corporation (Oracle Corporation)
• Original Assignee: Oracle International Corporation (Oracle Corporation)
• Inventors: Stamos, James W., Chatterjee, Debashish, Chidambaran, Lakshminarayanan
• Granted Patent: US 7,552,171 B2
• Time in Patent Office: Days
• US Class Current: 709/226
• CPC Class Codes: G06F 9/5088 involving task migration

Abstract

An approach migrates database sessions from a database server on which performance and resource availability requirements are not being met to another database server. Because workload is removed from the database server not meeting these requirements, improved performance there is promoted. Furthermore, the approach switches the users in a way that attempts to keep performance and resource availability levels at the database server receiving the database sessions within goals for performance and resource availability.

Claims

1. A method for balancing use of resources between nodes of a multi-node system, the method comprising the steps of:

detecting a service-level violation on a first node from a set of nodes in the multi-node system;

wherein the set of nodes hosts a multi-node server that includes a first server instance on the first node;

determining a first number of sessions running on the first server instance to migrate to a second server instance of the multi-node server on a second node of said set of nodes;

migrating at least a portion of said first number of sessions to said second server instance;

determining whether said service-level violation on the first node persists; and

if said service-level violation persists, repeating the step of determining a first number of instances and the step of migrating.

2. The method of claim 1, wherein the step of determining includes calculating a particular number of sessions estimated to avoid causing a service-level violation on the second node by migrating said particular number of sessions to said second server instance.

3. The method of claim 1, wherein the step of determining further includes dampening the particular number of sessions determined.

4. The method of claim 1, wherein the step of determining includes calculating a particular number of sessions estimated to avoid causing a service-level violation on the second node by migrating said particular number of sessions to said second server instance.

5. The method of claim 1, wherein the step of determining includes calculating a particular number of sessions estimated to reduce work performed by the first node sufficiently to eliminate said service-level violation.

6. The method of claim 1, wherein the step of determining a first number of sessions running on the first server instance to migrate is based on a set of one or more performance measures that includes CPU utilization.

7. The method of claim 1, wherein the step of determining a first number of sessions is based on:

S as a percent CPU utilization of the first node,

T as a percent CPU utilization on the second node, and

A as an average percent utilization of a set of sessions on the first node.

8. The method of claim 7, wherein the step of determining a first number includes calculating a particular number as N, where

$N=\frac{S-T}{2*A}.$

9. The method of claim 8, wherein the step of calculating a particular number is performed if a threshold of CPU utilization is greater than S and S is greater than T.

10. The method of claim 7, wherein:

X is a threshold of CPU utilization; and

the step of determining a first number includes calculating a particular number as N,

$\mathrm{wherein}<br><br>\text{\hspace{1em}<br><br>}<br><br>N=\frac{\mathrm{Minimum}<br><br>\left(S-X,X-T\right)}{A}.$

11. The method of claim 10, wherein the step of calculating a particular number is performed if S is greater than X and X is greater than T.

12. The method of claim 7, wherein:

X is a threshold of CPU utilization; and

the step of determining a first number includes calculating a particular number as N,

$N=\frac{X-T}{A}.$

13. The method of claim 12, wherein the step of calculating the first number is performed if T is greater than S and X is greater than T.

14. The method of claim 13, wherein:

a plurality of services are hosted by one or more server instances of said multi-node server, wherein each service of said plurality of services is work of particular type;

the steps further include;

selecting a particular service of said plurality of services;

selecting a server instance hosting said particular service on a node different than said first node; and

determining whether one or more sessions can be migrated to said server instance.

15. The method of claim 14, wherein the step of determining whether one or more sessions can be migrated includes said determining a first number.

16. The method of claim 1, wherein:

the multi-node server is a database server managing access to a database;

said first server instance manages access to said database; and

said second server instance manages access to said database.

17. The method of claim 1, wherein:

the set of nodes is a cluster with each node having shared access to a non-volatile storage; and

each node of said cluster accesses a data item stored on said non-volatile storage without requiring another node access said data item on said each node'"'"'s behalf.

18. The method of claim 1, wherein the step of determining a first number of sessions is based on:

a plurality of performance grades that each indicate relative performance that can be achieved on said first node and said second node; and

performance deltas, wherein each performance delta is associated with a certain performance grade of said plurality of performance grades and indicates a change to the certain performance grade caused by adding work to the node associated with the performance grade.

19. The method of claim 18, wherein the step of determining further includes dampening the number of sessions determined.

20. The method of claim 18, wherein the steps further include:

receiving a first version of performance grades of said plurality of performance grades before performing the step of determining a first number;

after the step of migrating, receiving a second version of said plurality of performance grades as a revised version of said first version of performance grades; and

wherein the step of repeating includes determining a first number based on the second version of performance of performance grades.

21. The method of claim 18, wherein each performance grade of said plurality of performance grades includes data indicating at least one status of a set of statuses, the set of statuses including:

a BLOCKED status associated with a node, and

a VIOLATING status associated with a node; and

wherein the step of migrating includes foregoing migrating any sessions based on said at least one status.

22. The method of claim 18, wherein

g₁ is a performance grade for the first node;

g₂ is a performance grade for the second node;

d₁ is a performance grade delta for the first node;

d₂ is a performance grade delta on the second node; and

the step of determining a first number includes calculating a particular number as N, where

N=(g₁−

g₂)/(d₁+d₂).