Resource-aware application scheduling

• US 20090165004A1
• Filed: 12/21/2007
• Published: 06/25/2009
• Est. Priority Date: 12/21/2007
• Status: Abandoned Application

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: Newell, Don K., Li, Tong N., Apparao, Padmashree, Moses, Jaideep, Iyer, Ravishankar, Makineni, Srihari, Illikkal, Ramesh, Zhao, Li, Hahn, Scott
• Application Number:
  US12/004,756
• Publication Number:
  US 20090165004A1
• Time in Patent Office: Days
• US Class Current: 718/102
• CPC Class Codes: G06F 2209/507 Low-level
• CPC Class Codes: G06F 2209/508 Monitor
• CPC Class Codes: G06F 9/505 considering the load

Abstract

In one embodiment, a method provides capturing resource monitoring information for a plurality of applications; accessing the resource monitoring information; and scheduling at least one of the plurality of applications on a selected processing core of a plurality of processing cores based, at least in part, on the resource monitoring information.

Claims

1. A method to schedule applications, comprising:

capturing resource monitoring information for a plurality of applications;

accessing the resource monitoring information; and

scheduling at least one of the plurality of applications on a selected processing core of a plurality of processing cores based, at least in part, on the resource monitoring information.

2. The method of claim 1, wherein the resource monitoring information comprises, for any given one of the plurality of applications, resource usage information.

3. The method of claim 2, wherein resource usage information comprises the application'"'"'s occupancy of a given shared cache amongst a plurality of shared caches.

4. The method of claim 3, wherein said scheduling at least one of the plurality of applications on a selected processing core based, at least in part, on the resource monitoring information comprises scheduling the application on a processing core that is connected to one of the plurality of shared caches having a high cache occupancy by the application.

5. The method of claim 4, wherein said scheduling is performed on a per-core task queue.

6. The method of claim 4 wherein said scheduling is performed on a global shared cache basis.

7. The method of claim 1 additionally comprising classifying the plurality of applications based on the resource monitoring information.

8. The method of claim 7, wherein the resource monitoring information comprises resource usage information and contention information.

9. The method of claim 8, wherein said classifying the plurality of applications based on the resource monitoring information comprises classifying each of the applications into one of a vulnerable application, a destructive application, and a neutral application based on a combination of resource usage information and contention information.

10. The method of claim 9, wherein said scheduling at least one of the plurality of applications on a selected processing core based, at least in part, on the resource monitoring information comprises:

pairing applications without pairing a destructive application with a vulnerable application; and

scheduling the paired applications on one of the plurality of processing cores.

11. An apparatus to schedule applications, comprising:

a capture module having a monitoring module to monitor resource monitoring information for a plurality of applications; and

a scheduling module to;

use the monitored resource monitoring information; and

schedule at least one of the plurality of applications on a selected processing core of a plurality of processing cores based, at least in part, on the resource monitoring information.

12. The apparatus of claim 11, said capture module additionally comprising a classification module to classify the plurality of applications based on the resource monitoring information.

13. The apparatus of claim 12, wherein said classification module additionally classifies each of the applications into one of a vulnerable application, a destructive application, and a neutral application.

14. The apparatus of claim 11, wherein said scheduling module additionally:

pairs applications without pairing a destructive application with a vulnerable application; and

schedules the paired applications on one of the plurality of processing cores.

15. The apparatus of claim 14, wherein said scheduling module comprises a load balancer to pair applications and schedule the paired applications.

16. An article of manufacture having stored thereon instructions, the instructions when executed by a machine, result in the following:

capturing resource monitoring information for a plurality of applications;

accessing the resource monitoring information; and

scheduling at least one of the plurality of applications on a selected processing core of a plurality of processing cores based, at least in part, on the resource monitoring information.

17. The article of claim 16, wherein the resource monitoring information comprises, for any given one of the plurality of applications, resource usage information.

18. The article of claim 17, wherein resource usage information comprises the application'"'"'s occupancy of a given shared cache amongst a plurality of shared caches.

19. The article of claim 18, wherein said scheduling at least one of the plurality of applications on a selected processing core based, at least in part, on the resource monitoring information comprises scheduling the application on a processing core that is connected to one of the plurality of shared caches having a high cache occupancy by the application.

20. The article of claim 19, wherein said scheduling is performed on a per-core task queue.

21. The article of claim 19, wherein said scheduling is performed on a global shared cache basis.

22. The article of claim 16 additionally comprising classifying the plurality of applications based on the resource monitoring information.

23. The article of claim 22, wherein the resource monitoring information comprises resource usage information and contention information.

24. The article of claim 23, wherein said classifying the plurality of applications based on the resource monitoring information comprises classifying each of the applications into one of a vulnerable application, a destructive application, and a neutral application based on a combination of resource usage information and contention information.

25. The article of claim 24, wherein said scheduling at least one of the plurality of applications on a selected processing core based, at least in part, on the resource monitoring information comprises:

pairing applications without pairing a destructive application with a vulnerable application; and

scheduling the paired applications on one of the plurality of processing cores.