Method and system for on-line performance modeling using inference for real production it systems

• US 20070168494A1
• Filed: 12/22/2005
• Published: 07/19/2007
• Est. Priority Date: 12/22/2005
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Xia, Honghui, Liu, Zhen, Zhang, Li, Zhang, Fan
• Granted Patent: US 7,739,099 B2
• Time in Patent Office: Days
• US Class Current: 709/224
• CPC Class Codes: G06Q 10/06 Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
• CPC Class Codes: Y02P 90/80 Management or planning

Abstract

A system and method for performance modeling for an information technology (IT) system having a server(s) for performing a number of types of transactions includes receiving data for system topology and transaction flows and receiving performance measurement data for the IT system. The measurement data is clustered into multiple regimes based on similarities. Service demand and network delay parameters may be inferred based on clustered data.

Claims

1. A method for performance modeling for an information technology (IT) system having one or more servers for performing a number of types of transactions, comprising:

receiving data for system topology and transaction flows;

receiving performance measurement data for the IT system;

clustering the performance measurement data into multiple regimes based on similarities; and

inferring service demand and network delay parameters based on the clustered performance measurement data.

2. The method as recited in claim 1, wherein receiving data for system topology and transaction flows includes receiving at least one modeled queue corresponding to at least one component of the system.

3. The method as recited in claim 1, wherein receiving performance measurement data includes receiving end-to-end response times of transactions carried out by the production IT system under a plurality of different scenarios.

4. The method as recited in claim 3, wherein the scenarios include transactions from multiple geographical locations, and/or at multiple times of the day or continuously.

5. The method as recited in claim 1, wherein receiving performance measurement data includes receiving online server utilization of the IT system continuously or at multiple times of the day.

6. The method as recited in claim 1, wherein inferring service demand and network delay parameters based on the clustered performance measurement data includes inferring service times of transactions handled by the one or more servers and network delay parameters from multiple geographical locations based on the clustered multiple regime performance measurement data.

7. The method as recited in claim 1, wherein inferring service demand and network delay parameters based on the clustered data includes:

deriving equations for the performance measurement data;

solving the equations to find at least one value corresponding to a service time for a transaction handled by one of the servers; and

selecting an optimum value of service time.

8. The method as recited in claim 7, wherein the step of selecting an optimum value comprises:

solving a quadratic program formulated with the clustered multiple regime performance measurement data;

comparing computed performance data with the received measurement performance data; and

selecting the value of a service time corresponding to the computed performance data being closest to the performance measurement data.

9. The method as recited in claim 1, wherein the step of inferring includes:

creating a stochastic multiple regime queuing network model of the IT system;

obtaining from the stochastic model at least one value corresponding to service time for a transaction handled by a server; and

searching for an optimum value of service time.

10. A computer program product for modeling for an information technology (IT) system having one or more servers for performing a number of types of transactions, comprising a computer useable medium having a computer readable program, wherein the computer readable program when executed on a computer causes the computer to perform:

receiving data for system topology and transaction flows;

receiving performance measurement data for the IT system;

clustering the performance measurement data into multiple regimes based on similarities; and

inferring service demand and network delay parameters based on the clustered performance measurement data.

11. The computer program product as recited in claim 10, wherein receiving data for system topology and transaction flows includes receiving at least one modeled queue corresponding to at least one component of the system.

12. The computer program product as recited in claim 10, wherein receiving performance measurement data includes receiving end-to-end response times of transactions carried out by the production IT system under a plurality of different scenarios.

13. The computer program product as recited in claim 12, wherein the scenarios include transactions from multiple geographical locations, and/or at multiple times of the day or continuously.

14. The computer program product as recited in claim 10, wherein receiving performance measurement data includes receiving online server utilization of the IT system continuously or at multiple times of the day.

15. The computer program product as recited in claim 10, wherein inferring service demand and network delay parameters based on the clustered data includes inferring service times of transactions handled by the one or more servers and network delay parameters from multiple geographical locations based on the clustered multiple regime performance measurement data.

16. The computer program product as recited in claim 10, wherein inferring service demand and network delay parameters based on the clustered data, includes:

deriving equations for the performance measurement data;

solving the equations to find at least one value corresponding to a service time for a transaction handled by one of the servers; and

selecting an optimum value of service time.

17. The computer program product as recited in claim 16, wherein the step of selecting an optimum value comprises:

solving a quadratic program formulated with the clustered multiple regime measurement data;

comparing computed performance data with the received measurement performance data; and

selecting the value of a service time corresponding to the computed performance data being closest to the performance measurement data.

18. The computer program product as recited in claim 10, wherein the step of inferring includes:

creating a stochastic multiple regime queuing network model of the IT system;

obtaining from the stochastic model at least one value corresponding to service time for a transaction handled by a server; and

searching for an optimum value of service time.

19. A system for modeling for an information technology (IT) system having one or more servers for performing a number of types of transactions, comprising:

a first module which receives data for system topology, transaction flows and performance measurement data for the IT system to be modeled;

a clustering module which clusters the measurement data into multiple regimes based on similarities; and

an inferring module to infer service demand and network delay parameters based on clustered data to form a model to simulate operations and conditions of the IT system under a plurality of conditions.

20. The system as recited in claim 19, further comprising a module for performing performance prediction and optimization of performance by permitting a plurality of parameters to be altered to improve modeling of the IT system.

21. The system as recited in claim 19, further comprising a model validation module to refine a model based upon available data.