Calculation of sevice performance grades in a multi-node environment that hosts the services

• US 20050038800A1
• Filed: 08/12/2004
• Published: 02/17/2005
• 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., Chidambaran, Lakshminarayanan, Kantarjiev, Christopher A., Povinec, Peter
• Granted Patent: US 7,437,459 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: H04L 67/1008 based on parameters of servers, e.g. available memory or workload monitoring of computer activity G06F11/30
• CPC Class Codes: H04L 67/1031 Controlling of the operation of servers by a load balancer, e.g. adding or removing servers that serve requests

Abstract

An approach generates performance grades. A performance grade is a value or set of values that indicates the relative work performance offered by a resource, such as a server in a multi-node system, relative to other resources. Performance grades are used to dynamically allocate work within a multi-node computer system. For example, performance grades are used for connection-time balancing, and to direct a database connection request to a multi-node database instance that a performance grade indicates provides better performance. Performance grade deltas are generated that reflect the incremental change in performance grade for additional sessions or work. Performance deltas are used to direct incoming sessions, to more optimally balance workload, and to deal with logon storms where a large number of connect requests arrive in a very short time period.

Claims

1. A method used for managing resources in a multi-node system, the method comprising the steps of:

a plurality of performance grades that each indicate relative performance that can be realized for categories of work performed by a plurality of servers;

generating

wherein each server of said plurality of servers runs on a node in said multi-node system;

wherein each performance grade of said plurality of performance grades indicates, for a particular category of work, relative performance that may be realized on a server relative to performance that may be realized on other servers of said plurality of servers; and

adjusting allocation of servers between said categories of work based on said performance grades.

2. The method of claim 1, further including generating 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 adjusting allocation of work to the server associated with the performance grade.

3. The method of claim 2, wherein the step of generating performance deltas includes generating a quotient of:

the difference between a revised performance grade for the particular category of work and a previous value for the revised performance grade, divided by the difference of a number of sessions associated with the revised performance grade and a previous value of a number sessions for the particular category of work.

4. The method of claim 2, wherein adjusting allocation of servers between said categories of work includes:

adjusting allocation of a first server of said plurality of servers for a first category of work;

wherein said first server is associated with a first performance grade and a first performance delta; and

in response to allocating said first server, changing the first performance grade based on the first performance delta.

5. The method of claim 4, wherein:

adjusting allocation of a first server of said plurality of servers for a first category of work includes establishing a new session on said first server; and

changing the first performance grade includes changing the first performance grade to reflect the establishment of said new session.

6. The method of claim 4, wherein:

adjusting allocation of a first server of said plurality of servers for a first category of work includes causing termination of a certain session on said first server; and

changing the first performance grade includes changing the first performance grade to reflect the termination of said certain session.

7. The method of claim 2, wherein:

the step of generating performance grades is based on a mathematical function; and

the step of generating performance deltas is based on one or more derivatives of said mathematical function.

8. The method of claim 2, wherein the derivative is calculated based on performance metrics generated for said categories of work.

9. The method of claim 1, wherein the step of generating a plurality of performance grades is based on one or more of the following:

a value representing a proportion of CPU usage, on a certain server of said plurality of servers, realized for a certain category of work per each request of a subset of requests associated with said certain category of work; and

a value representing an amount of time that elapses for said each request.

10. The method of claim 9, wherein the step of generating performance grades is based on a mathematical function X/Y, wherein:

X represents said proportion of CPU usage; and

Y represents said amount of time that elapses for said each request.

11. The method of claim 10, further including generating performance deltas, wherein:

each performance delta of said performance deltas indicates a change to a performance grade caused by allocating work to the server associated with the performance grade; and

the step of generating performance deltas is based on a derivative of said mathematical function.

12. The method of claim 1, wherein:

a first system component performs the step of generating a plurality of performance grades;

the steps further include said first system component transmitting to a second system component said plurality of performance grades; and

said second system component performs the step of adjusting allocation of servers.

13. The method of claim 12, wherein:

the steps further include said first system component transmitting to said second system component information about a violation of one or more service-level agreements for said categories of work that occurs on said plurality of servers; and

said second system component performs the step of adjusting allocation of servers based on said information.

14. The method of claim 12, wherein:

the steps further include said first system component transmitting to said second system component certain information about a certain category of work of said categories that is hosted on said plurality of servers; and

said second system component performs the step of adjusting allocation of servers based on said certain information about a certain category of work.

15. The method of claim 14, wherein said certain information includes information about at least one of the following:

that a particular server is no longer allocated to host the certain category of work; and

that a particular performance grade has not been generated for the certain category of work hosted on a particular server.

16. The method of claim 1, wherein:

a first system component performs the step of generating a plurality of performance grades, wherein the step of generating a plurality of performance grades includes generating a first version of performance grades of the said plurality of performance grades and a second version of performance grades of said plurality of performance grades;

the steps further include generating a plurality of performance grades that each indicate relative performance that can be realized for categories of work performed by a plurality of servers;

the steps further include said first system component transmitting to a second system component said plurality of performance grades;

said second system component performs the step of adjusting allocation of servers; and

the steps further include;

said second system component receiving said first version of performance grades and storing a copy of said first version of performance grades,

subsequent and separate to receiving said first version of performance grades, said second system component receiving said second version of performance grades,

said second system component receiving performance deltas, wherein each performance delta is associated with a certain performance grade of said first version of performance grades and indicates a change to the certain performance grade caused by allocating work to the server associated with the performance grade, and

in response to adjusting allocation of servers before receiving said second version of performance grades, said second system component adjusting said copy of said first version of performance grades based on said performance deltas.