Translating service level objectives to system metrics

• US 8,756,307 B1
• Filed: 07/30/2007
• Issued: 06/17/2014
• Est. Priority Date: 07/30/2007
• Status: Active Grant

• Current Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Chen, Yuan, Milojicic, Dejan S., Iyer, Subramoniam N., Sahai, Akhil, Liu, Xue
• Primary Examiner(s): WASEL, MOHAMED A
• Application Number:
  US11/830,586
• Time in Patent Office: 2,514 Days
• Field of Search: 709223-226
• US Class Current: 709/224
• CPC Class Codes: G06F 11/3414 Workload generation, e.g. scripts, playback
• CPC Class Codes: G06F 11/3442 for planning or managing the needed capacity
• CPC Class Codes: G06F 11/3447 Performance evaluation by modeling

Abstract

Service level objectives for a multi-tier system are translated to low-level system metrics by determining component profiles. The component profiles include performance parameters calculated as a function of applied resources for each of a plurality of components of the multi-tier system. Performance models including service level parameters are also determined. The performance models are determined as a function of the performance parameters and workload characteristics. The service level objectives are translated to the low-level system metrics using the performance models.

Claims

1. A method comprising:

determining performance parameters as a function of applied resources for each of a plurality of components of a multi-tier system;

modeling the plurality of components as a closed queuing network, wherein at least one of the plurality of components is modeled as a multiple station queuing center representative of a multi-threaded server architecture;

modeling service level parameters, wherein the modeling of each service level parameter is determined as a function of at least one of the performance parameters and workload characteristics; and

translating service level objectives for the multi-tier system to low-level system metrics using the models of the service level parameters.

2. The method of claim 1, wherein translating service level objectives to low-level system metrics using the models of the service level parameters comprises:

calculating a quantity of applied resources for each component of the multi-tier system using the models of the service level parameters, wherein the calculated quantity of applied resources are sufficient to ensure that each service level objective is satisfied.

3. The method of claim 2, wherein calculating a quantity of applied resources for each component of the multi-tier system using the models of the service level parameters comprises:

incrementally varying a quantity of applied resources to determine each of the performance parameters for each incremental variation;

applying the performance parameters for each incremental variation to the models of the service level parameters until the service level objectives are satisfied.

4. The method of claim 3, wherein applying the performance parameters for each incremental variation to the models of the service level parameters until the service level objectives are satisfied comprises:

applying the performance parameters for each incremental variation to the models of the service level parameters until the service level objectives are satisfied and an objective function for allocating resources to the multi-tier system is met.

5. The method of claim 1, wherein the service level objectives are values for the service level parameters.

6. The method of claim 1, wherein translating service level objectives for the multi-tier system to low-level system metrics using the models of the service level parameters comprises:

determining configuration parameters for the multi-tier system such that the service level objectives are satisfied.

7. The method of claim 1, wherein the service level parameters includes system response time and a throughput.

8. The method of claim 1, wherein modeling the plurality of components comprises:

modeling a first component of the plurality of components as a multiple station queuing center by approximating the multiple station queuing center with first and second queues processed in tandem to one another.

9. The method of claim 8, wherein approximating the multiple station queuing center comprises:

modeling the first queue as a single-station queue that has a modeled service demand equal to a prorated share of a total service demand processed by the multiple station queuing center.

10. The method of claim 9, wherein the modeled service demand equals D/m where D is the total service demand processed by the multiple station queuing center and m is a number of stations in the multiple station queuing center.

11. The method of claim 8, wherein approximating the multiple station queuing center comprises:

modeling the second queue as a single-station queue that has a modeled delay proportional to a total service demand at the queuing center.

12. The method of claim 8, wherein the modeled delay equals D×

(m−

1)/m where D is the total service demand processed by the multiple station queuing center and m is a number of stations in the multiple station queuing center.

13. The method of claim 1, further comprising:

determining an allocation of resources based on the low-level system metrics, wherein the allocation of resources includes allocating resources from a pool of resources in a resource-on-demand system for virtual machines running applications of the components in the multi-tier system.

14. A system including a computer storage medium operable to translate service level objectives to low-level system metrics, the system comprising:

a plurality of component profiles, each component profile corresponding to a component of a plurality of components in a multi-tier system, and each component profile includes a performance parameter calculated as a function of applied resources to a component of the multi-tier system, wherein the plurality of components are modeled as a closed queuing network and at least one of the plurality of components is modeled as a multiple station queuing center representative of a multi-threaded server architecture;

a plurality of performance models, wherein each performance model is determined as a function of at least one of the performance parameters and workload characteristics; and

a controller configured to translate service level objectives for the multi-tier system to low-level system metrics using the performance models.

15. The system of claim 14, wherein the controller is configured to calculate a quantity of applied resources for each component of the multi-tier system using the performance models, wherein the calculated quantity of applied resources are sufficient to ensure that each service level objective is satisfied.

16. The system of claim 15, wherein the controller is configured to calculate a quantity of applied resources for each component of the multi-tier system based on an objective function for allocating resources to the multi-tier system.

17. A computer system comprising:

data storage storing a plurality of component profiles, each component profile corresponding to a component of a plurality of components in a multi-tier system, and each component profile includes a performance parameter calculated as a function of applied resources to a component of the multi-tier system

wherein the plurality of components are modeled as a closed queuing network and at least one of the plurality of components is modeled as a multiple station queuing center representative of a multi-threaded server architecture, and

wherein the data storage stores a plurality of performance models, wherein each performance model is determined as a function of at least one of the performance parameters and workload characteristics; and

a computer readable medium including instructions that when executed by a processor translate service level objectives for the multi-tier system to low-level system metrics using the performance models.

18. The computer system of claim 17, wherein the instructions when executed calculate a quantity of applied resources for each component of the multi-tier system using the performance models, wherein the calculated quantity of applied resources are sufficient to ensure that each service level objective is satisfied.