Use of Metrics to Control Throttling and Swapping in a Message Processing

• US 20080276238A1
• Filed: 07/14/2008
• Published: 11/06/2008
• Est. Priority Date: 11/14/2003
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Smith, Kevin Bowen, Sriram, Balasubramanian, Maybee, Paul, Saha, Sanjib, Graber, Lee, Levanoni, Yossi, Mehta, Bimal Kumar, Musayev, Eldar Azerovich
• Granted Patent: US 9,471,392 B2
• Time in Patent Office: Days
• US Class Current: 718/100
• CPC Class Codes: G06F 9/4881 Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
• CPC Class Codes: G06F 9/5083 Techniques for rebalancing the load in a distributed system

Abstract

A system and method of using metrics to control throttling and swapping in a message processing system is provided. A workload status of a message processing system is determined, and the system polls for a new message according to the workload status. The message processing system identifies a blocked instance and calculates an expected idle time for the blocked instance. The system dehydrates the blocked instance if the expected idle time exceeds a predetermined threshold.

Claims

1. A system for improving the efficiency of a message processing system, the system comprising a processor configured to:

utilize performance metrics to determine the workload of the message processing system and to determine whether the message processing system is overburdened with messages;

restrict the number and/or type of new messages, wherein the system is able to help the message processing system to process its existing work if the system determines the message processing system to be overburdened; and

utilize performance metrics to predict how long an idle process will remain idle, to enable a determination as to whether the system should move the process out of active memory and into secondary storage, and to improve system performance.

2. The system of claim 1, wherein performance metrics are based on memory consumed by active message processing.

3. The system of claim 1, wherein performance metrics are based on processing power used for active message processing.

4. The system of claim 1, wherein performance metrics are based an idle time of a past process.

5. The system of claim 1, wherein performance metrics are assigned according to a predetermined criterion if no performance metrics are available.

6. The system of claim 1, wherein the system determines whether the predicted duration of idle time exceeds a predetermined threshold.

7. The system of claim 6, wherein the predetermined threshold is based on a prior idle process.

8. A method for improving the efficiency of a message processing system, the method comprising:

using performance metrics to determine the workload of the message processing system and to determine whether the message processing system is overburdened with messages;

restricting the number and/or type of new messages, wherein the system is able to help the message processing system to process its existing work if the system determines the message processing system to be overburdened; and

using performance metrics to predict how long an idle process will remain idle.

9. The method of claim 8, wherein performance metrics are based on memory consumed by active message processing.

10. The method of claim 8, wherein performance metrics are based on processing power used for active message processing.

11. The method of claim 8, further comprising using the prediction of idle time to enable a determination as to whether the system should move the process out of active memory and into secondary storage, thereby improving system performance.

12. The method of claim 8, wherein performance metrics are based an idle time of a past process.

13. The method of claim 8, wherein performance metrics are assigned according to a predetermined criterion if no performance metrics are available.

14. The method of claim 8, wherein the system determines whether the predicted duration of idle time exceeds a predetermined threshold.

15. The method of claim 14, wherein the predetermined threshold is based on a prior idle process.

16. A system for improving the efficiency of a message processing system, the system comprising a processor configured to:

determine a workload of a message processing system by accessing performance data regarding the message processing system, and determining, using the performance data, the workload with respect to a system operating parameter;

poll for a new message at a frequency according to the workload status, wherein the frequency is inversely proportional to the workload and, if the workload is above a predetermined limit, polling only for a new non-activation message;

identify a blocked instance being processed by the message processing system;

calculate an expected idle time for the blocked instance by;

accessing performance data for the message processing system;

determining a length of time the blocked instance has been idle; and

generating the expected idle time based on the performance data and length of

time the blocked instance has been idle;

dehydrate the blocked instance if the expected idle time exceeds a predetermined threshold;

update the workload according to the dehydration of the instance; and

update the threshold according to the workload.

17. The system of claim 16, wherein the accessed performance data comprises at least one of:

memory usage, processor power in use by the message processing system, or assignments according to a predetermined criterion if no performance data is accessible.

18. The system of claim 16, wherein the polling step is carried out at one of a first or second frequencies, wherein the first frequency is greater than the second frequency.

19. The method of claim 18, wherein the polling step further comprises polling only for a new non-activation message.

20. The method of claim 16, wherein the blocked instance is a first instance, and the performance data comprises a recorded idle time of a second instance.