Comparative performance modeling for distributed object oriented applications

• US 5,881,268 A
• Filed: 09/30/1996
• Issued: 03/09/1999
• Est. Priority Date: 03/14/1996
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Qin, Bin, Thompson, Brighid Anne, McDonald, Richard Denison, Rass, Anita Krista
• Primary Examiner(s): Teska, Kevin J.
• Assistant Examiner(s): Fiul, Dan
• Application Number:
  US08/723,180
• Time in Patent Office: 890 Days
• Field of Search: 395/500, 395/183.07, 395/800.28, 701/103, 379/201
• US Class Current: 703/21
• CPC Class Codes: G05B 17/02 electric

Abstract

A performance modeling tool and method permitting the user to define the elements of a distributed system (hosts, networks and response times), and examine the effect on performance of different distributions of application processes over the system at an early stage in application design. Once a user has defined a performance scenario, it is saved to a data model as a number of interdependent persistent objects that show the distribution of the application for a particular performance scenario from different views. Multiple alternates of each object can be stored. The user can construct different performance scenarios for analysis from the stored objects. Analysis can include performance simulation from which the user can obtain performance projections for an application process or several application processes over different distributions of the performance worload.

Claims

1. A mechanism for providing performance projections for designs of an application over a distributed system, comprising:

means for defining different parameters of the application and the distributed system as a plurality of views including a partition, scenario, topology and workload views;

means for separately storing the defined parameters as persistent objects for each of said views;

means for modifying the stored persistent objects within each of said views; and

means for providing analyses of user selected groupings of the stored persistent objects that define different performance scenarios based upon each of said views.

2. A mechanism, according to claim 1, wherein the means for defining different parameters of the application and the distributed system comprise a graphical user interface adapted to receive input from a user and schematically display distributed objects and networks from the distributed system, and interaction links between the distributed objects from the application, and wherein the means for storing the defined parameters as persistent objects comprise means to store as separate objects alternate distributions of the distributed objects and alternate interaction links between the distributed objects.

3. A mechanism, according to claim 2, wherein the means for modifying the stored persistent objects comprise means for recalling the stored persistent objects to the graphical user interface.

4. A mechanism, according to claim 1, wherein the means for providing analyses, comprises:

means for selecting a set of parameters defining a distributed system from the stored persistent objects;

means for simulating an application process over the set of parameters; and

means for providing performance results of the simulation.

5. A mechanism, according to claim 1, wherein the means for defining different parameters of the application and the distributed system comprise a graphical user interface having a plurality of views of the application and the distributed system, at least one view representing a partition view adapted to receive input from a user and schematically display a plurality of distributed objects and interaction links between the distributed objects.

6. A mechanism, according to claim 5, wherein the partition view further comprises means for defining middleware communication values in the interaction links.

7. A mechanism, according to claim 5, wherein the plurality of views further comprises at least one view representing a scenario view adapted to receive input from the user and schematically display interactions between distributed objects for performing an application process.

8. A mechanism, according to claim 1, wherein the means for defining different parameters of the application and the distributed system comprise a graphical user interface having a plurality of views of the application and the distributed system, at least one view representing a topology view adapted to receive input and display representations of a plurality of hosts and networks between the hosts in the distributed system.

9. A mechanism, according to claim 7, wherein the plurality of views further comprises at least one view representing a topology view adapted to receive input and display representations of a plurality of hosts and networks between the hosts in the distributed system.

10. A mechanism, according to claim 9, wherein the plurality of views further comprises at least one view representing a workload view adapted to receive input and display arrival rates for the application process.

11. A method, implemented in a computer, for determining a performance-based distribution of application processes over a distributed system having a plurality of linked processors and including middleware connections between distributed processors, the method comprising the steps of:

i) selecting parameters defining

a) hosts, networks and connections in the distributed system,

b) interactions between the hosts for executing at least one application process; and

c) storing said selected parameters as a plurality of separate views including a partition, scenario, topology and workload views;

ii) executing a simulation of said at least one application process according to the selected parameters in said partition, scenario, topology and workload views; and

iii) providing a performance analysis of the simulation based upon said views.

12. A method, according to claim 11, further comprising the steps of:

receiving input to define alternate parameters for at least one of the hosts, networks, connections or interactions between hosts; and

saving said alternate parameters as separate data objects.

13. A method, according to claim 12, wherein the step of selecting parameters comprises selecting parameters from the saved data objects.

14. A method, according to claim 11, further comprising the step of repeating steps ii) and iii) for alternate selections of parameters.

15. A method, according to claim 12, wherein the step of receiving input to define alternate parameters comprises receiving input to define at least:

a partition view having partitions, objects within the partitions, associations between the objects and middleware defaults;

a scenario view having interactions between the objects and resource demands on the distributed system;

a topology view having hosts, networks, connections between the hosts and the networks, and processes on the hosts; and

a workload view having values for arrival rates for external events.

16. A method, according to claim 15, wherein the workload view is defined as further comprising required response time for external service requests.

17. A method, according to claim 15, wherein the step of receiving input to define alternate parameters comprises receiving input to define at least one alternate of one of said views.

18. A computer program product having a computer readable medium for use with a graphic display device in a computer, said computer program product comprising:

computer readable program code means embodied in said medium for providing performance projections for designs of an application over a distributed system, said computer program product having;

computer readable code means for causing a computer to define different parameters of the application and the distributed system as a plurality of views including a partition, scenario, topology and workload views in response to user input through the graphics display device;

computer readable code means for causing the computer to separately store the defined parameters as persistent objects for each of said views;

computer readable code means for modifying the stored persistent objects within each of said views in response to user input through the graphics display device; and

computer readable code means for causing the computer to provide analyses of groupings of the stored persistent objects, as selected by user input, said groupings defining different performance scenarios based upon each of said views.

19. A computer program product, according to claim 18, wherein the computer readable program code means for causing the computer to provide analyses also causes the computer to:

select a set of parameters defining a distributed system from the stored persistent objects;

simulate an application process over the set of parameters; and

provide performance results of the simulation.

20. A program storage device readable by a machine, tangibly embodying a program of instructions executable by the machine to perform method steps for determining a performance-based distribution of application processes over a distributed system having a plurality of linked processors and including middleware connections between distributed processors, said method steps comprising:

i) selecting parameters defining

a) hosts, networks and connections in the distributed system,

b) interactions between the hosts for executing at least one application process; and

c) storing said selected parameters as a plurality of separate views including a partition, scenario, topology and workload views;

ii) executing a simulation of said at least one application process according to the selected parameters in said partition, scenario, topology and workload views; and

iii) providing a performance analysis of the simulation based upon said views.

21. A program storage device readable by a machine, tangibly embodying a program of instructions executable by the machine to perform method steps for determining a performance-based distribution of application processes over a distributed system having a plurality of linked processors and including middleware connections between distributed processors, said method steps comprising:

i) receiving input to define alternate parameters to define;

a partition view having partitions, objects within the partitions, associations between the objects and middleware defaults;

a scenario view having interactions between the objects and resource demands on the distributed system;

a topology view having hosts, networks, connections between the hosts and the networks, and processes on the hosts; and

a workload view having values for arrival rates for external events.

ii) saving said alternate parameters as separate data objects;

iii) executing a simulation of said at least one application process according to the selected parameters; and

iv) providing a performance analysis of the simulation.