Automatic generation of hybrid performance models

• US 7,805,496 B2
• Filed: 05/10/2005
• Issued: 09/28/2010
• Est. Priority Date: 05/10/2005
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Sela, Aviad, Gilat, Dagan, Landau, Ariel, Aiber, Sarel, Razinkov, Natalia, Wasserkrug, Segev
• Primary Examiner(s): Etienne; Ario
• Assistant Examiner(s): Halim; Sahera
• Application Number:
  US11/125,523
• Publication Number:
  US 20070006177A1
• Time in Patent Office: 1,967 Days
• Field of Search: 709/217, 709/248, 709/216, 702/186
• US Class Current: 709/217
• CPC Class Codes: G06Q 10/10 Office automation; Time management

Abstract

A method for simulating a computer system includes defining a set of building blocks including models of components of the computer system. The set of building blocks is interconnected to produce a topological model of the computer system. A client transaction model is derived based on historical data, for generating simulated client requests to be processed by the topological model. A resource requirement model is produced based on the topological model and on the historical data, the resource requirement model simulating a resource utilization of the components of the computer system responsively to the generated client requests. A performance metric of the computer system is estimated by simulating the computer system using the simulated client requests and the resource requirement model.

Claims

1. A method for simulating a computer system, comprising:

defining a set of building blocks comprising models of components of the computer system;

interconnecting the set of building blocks to produce a topological model of the computer system;

deriving a client transaction model based on historical data, for generating simulated client requests to be processed by the topological model;

producing a resource requirement model based on the topological model and on the historical data, the resource requirement model simulating a resource utilization of the components of the computer system responsively to the generated client requests; and

estimating a performance metric of the computer system by simulating the computer system using the simulated client requests and the resource requirement model,

wherein deriving the client transaction model comprises deriving a user behavior model comprising one or more session types, which comprise characteristic sequences of client request types, and wherein deriving the user behavior model comprises;

defining simulated clients;

defining for each simulated client a session type frequency vector comprising relative frequencies of initiating the one or more session types by the simulated client; and

classifying the simulated client in a client cluster responsively to the session type frequency vector.

2. The method according to claim 1, wherein the computer system comprises an electronic-business (e-business) web-site.

3. The method according to claim 1, wherein defining and interconnecting the set of building blocks comprise receiving inputs from a planner using a graphical user interface (GUI) to construct the topological model.

4. The method according to claim 1, wherein defining the set of building blocks comprises defining core infrastructure models that simulate hardware and middleware components of the computer system, and modeling the infrastructure models using at least one of a state-machine model and a queuing network model.

5. The method according to claim 1, wherein defining the set of building blocks comprises, for some of the components, defining two or more alternative building blocks having different levels of complexity.

6. The method according to claim 1, wherein deriving the client transaction model comprises deriving a user attribute model that produces user attributes to be inserted into the generated simulated client requests.

7. The method according to claim 1, wherein producing the resource requirement model comprises producing a tier-level message breakdown model comprising sequences of tasks that are invoked by the respective client request types in the user behavior model.

8. The method according to claim 7, wherein producing the tier-level message breakdown model comprises aggregating, for each client request type in the user behavior model, a plurality of task sequences invoked by the client requests that correspond to the client request type, and representing the plurality of task sequences in terms of a probabilistic graph.

9. The method according to claim 7, wherein producing the resource requirement model comprises determining the resource utilization caused by the tasks in each of a plurality of tiers of the tier-level message breakdown model.

10. The method according to claim 1, wherein producing the resource requirement model comprises deriving a dynamic Bayesian network model responsively to the historic data.

11. The method according to claim 1, and comprising modifying at least one of the topological model, a business-related policy and an IT-related policy, so as to improve the estimated performance metric.

12. The method according to claim 11, wherein estimating the performance metric comprises automatically running a simulation, responsively to an event in the computer system, in order to evaluate a change in the performance metric due to modifying the at least one of the topological model, the business-related policy and the IT-related policy.

13. A method for performing an interactive analysis of a computer system to devise an information technology solution applicable to the computer system, the method comprising:

defining a set of building blocks comprising models of components of the computer system;

interconnecting the set of building blocks to produce a topological model of the computer system;

deriving a client transaction model based on historical data, for generating simulated client requests to be processed by the topological model;

producing a resource requirement model based on the topological model and on the historical data, the resource requirement model simulating a resource utilization of the components of the computer system responsively to the generated client requests; and

estimating a performance metric of the computer system by simulating the computer system using the simulated client requests and the resource requirement model,

wherein deriving the client transaction model comprises deriving a user behavior model comprising one or more session types, which comprise characteristic sequences of client request types, and wherein deriving the user behavior model comprises;

defining simulated clients;

defining for each simulated client a session type frequency vector comprising relative frequencies of initiating the one or more session types by the simulated client; and

classifying the simulated client in a client cluster responsively to the session type frequency vector.

14. The method according to claim 13, wherein the computer system comprises an electronic-business (e-business) web-site.

15. The method according to claim 13, wherein defining the set of building blocks comprises defining core infrastructure models that simulate hardware and middleware components of the computer system, and modeling the infrastructure models using at least one of a state-machine model and a queuing network model.

16. The method according to claim 13, wherein defining the set of building blocks comprises, for some of the components, defining two or more alternative building blocks having different levels of complexity.

17. The method according to claim 13, and comprising modifying at least one of the topological model, a business-related policy and an IT-related policy, so as to improve the estimated performance metric.