Optimal stress exerciser for computer servers

1. A method for selecting tests to stress test a computer system, comprising:

testing the computer system using a set of tests, wherein a given test includes a given load and a given cycling time selected from a range of cycling times;

for the given test, monitoring a stress metric in the computer system,

wherein the stress metric is determined from temperature samples measured at different locations in the computer system, and

wherein at least some of the temperature samples are multiplied by an associated weight when determining the stress metric; and

selecting at least one of the tests from the set of tests to stress test the computer system based on the monitored stress metric.

2. The method of claim 1 , wherein the given cycle time for the given load can include an idle time, during which the given load is not executed.

3. The method of claim 1 , wherein the given load includes an application.

4. The method of claim 3 , wherein the application is configured to stress a portion of the computer system.

5. The method of claim 4 , wherein the portion of the computer system includes one or more of the following: a processor, memory, an application-specific integrated circuit, an input/output interface, or a disk drive.

6. The method of claim 1 , wherein the selected test corresponds to the worst-case monitored stress metric.

7. The method of claim 1 , wherein the selected test is used during reliability or failure-analysis testing of the computer system.

8. The method of claim 1 , wherein tests are selected for different types of computer systems, and wherein a given selected test is subsequently used for a given type of computer system.

9. The method of claim 1 , wherein the stress metric includes thermal dynamics of the computer system.

10. The method of claim 1 , wherein the determination of the stress metric involves:

computing a length of a line between adjacent temperature samples, wherein the line includes a component that is proportionate to a difference between values of the adjacent temperature samples and a component that is proportionate to a time interval between the adjacent temperature samples; and

adding the computed length to a cumulative length variable, which is used as the stress metric.

11. The method of claim 10 , wherein the computed length is adjusted based on a function of the magnitude of the adjacent temperature samples.

12. The method of claim 11 , wherein the adjustment involves multiplying the length of the lines by an associated weight.

13. The method of claim 10 , wherein computing the length of the line between adjacent temperature samples involves computing √{square root over (|S 1 −S 2 | 2 +t 2 )}, where S 1 and S 2 are magnitudes of the adjacent temperature samples and t is the magnitude of the time interval between the adjacent temperature samples.

14. The method of claim 10 , wherein the temperature samples are measured at pre-determined time intervals.

15. A computer-program product for use in conjunction with a computer system, the computer program product comprising a computer-readable storage medium and a computer-program mechanism embedded therein for selecting tests to stress test a computer system, the computer-program mechanism including:

instructions for testing the computer system using a set of tests, wherein a given test includes a given load and a given cycling time selected from a range of cycling times;

for the given test, instructions for monitoring a stress metric in the computer system,

wherein the stress metric is determined from temperature samples measured at different locations in the computer system, and

wherein at least some of the temperature samples are multiplied by an associated weight when determining the stress metric; and

instructions for selecting at least one of the tests from the set of tests to stress test the computer system based on the monitored stress metric.

16. An apparatus, comprising:

a processor;

memory; and

a program module, wherein the program module is stored in the memory and configured to be executed by the processor, the program module including:

instructions for testing the computer system using a set of tests, wherein a given test includes a given load and a given cycling time selected from a range of cycling times;

for the given test, instructions for monitoring a stress metric in the computer system,

wherein the stress metric is determined from temperature samples measured at different locations in the computer system, and

wherein at least some of the temperature samples are multiplied by an associated weight when determining the stress metric; and

instructions for selecting at least one of the tests from the set of tests to stress test the computer system based on the monitored stress metric.