Correlation based adaptive system monitoring

1. A computer-implemented method of dynamic adaptive monitoring of a computer system over time to reduce redundant measurements in the computer system, comprising:

a server processing system identifying a group of performance parameters, from a multitude of performance parameters, for monitoring the computer system, including

the server processing system receiving performance indicators of a plurality of processes of the computer system,

the server processing system constructing a correlation matrix among the received performance indicators, wherein the correlation matrix is a table holding correlation values between elements of the computer system,

the server processing system using the correlation matrix to construct a time-varying correlation graph indicating defined, changing correlations between a multitude of performance parameters of the computer system, and

the server processing system mining said defined correlations indicated in the correlation graph to identify monitoring logic for monitoring the elements of the computer system, including selecting a group of performance parameters, from the multitude of performance parameters, for monitoring the elements of the computer system;

using the selected group of the performance parameters to monitor the elements of the computer system; and

the server processing system dynamically altering the group of performance parameters selected for monitoring the elements of the computer system as the time-varying correlation graph changes over time by selecting different groups of the performance parameters at different times for monitoring the elements of the computer system based on changes in the defined correlations indicated in the correlation graph to reduce redundant measurements in the computer system.

2. The method according to claim 1 , wherein:

the using the correlation graph further includes identifying a rate of change of the correlation graphs using a defined criteria; and

the using the parameters selected group of the performance includes using each of the selected groups of the performance parameters for monitoring the system for an associated time period,

and changing the lengths of the time periods based on the rate of change of the correlation graph.

3. The method according to claim 1 , wherein:

the using the correlation graphs further includes identifying a rate of change of the correlation graphs using a defined criteria; and

the changing the group of performance parameters includes dynamically changing the group of performance parameters based on the rate of change of the correlation graphs.

4. The method according to claim 1 , wherein the generating time varying correlation graphs includes:

during a first time period, monitoring a first plurality of said multitude of parameters of the system; and

generating one of the correlation graphs from said first plurality of said parameters.

5. The method according to claim 4 , wherein the using the correlation graphs includes:

analyzing said one of the correlation graphs to select a second plurality of said parameters of the system; and

using said second plurality of said parameters to monitor the system during a second time period.

6. The method according to claim 5 , wherein the second time period is after the first time period.

7. The method according to claim 1 , wherein the correlation graphs identify the ones of the multitude of parameters that have a predefined correlation with others of the multitude of parameters.

8. The method according to claim 1 , wherein each of the correlation graphs includes:

a multitude of nodes representing the multitude of parameters; and

a multitude of edges connecting the nodes together and representing the correlations between the multitude of parameters.

9. The method according to claim 1 , wherein the multitude of parameters include elements and processes of the system, and the using the correlation graphs includes querying the correlation graphs for one or more defined metrics.

10. The method according to claim 1 , wherein:

each of the correlation graphs includes a multitude of nodes representing the processes of the computer system, and a multitude of edges connecting the nodes together and denoting correlation values between the nodes of the correlation graphs;

the correlation graphs change over successive time periods; and

as the correlation graphs change over successive time periods, the using the correlation graphs includes:

using the changing correlation graphs to identify different groups of the parameters as monitoring parameters, wherein for each of the time periods, one of said different groups is identified as the monitoring parameters for said each time period; and

during each of the time periods, using the one of the different groups of monitoring parameters identified as the monitoring parameters for said each time period, to monitor the system during said each time period.

11. The method according to claim 1 , wherein the computer system includes specified hardware devices, and the method further comprises:

operating said specified hardware devices; and

using the selected group of parameters for monitoring the operating specified hardware devices.

12. The method according to claim 1 , wherein the using, by the processing system, said selected information obtained from the correlation graph includes using said selected information to determine which ones of the parameters to monitor at a higher rate and which ones of the parameters to monitor at a lower rate.

13. A method of correlation based, dynamic adaptive monitoring of a computer system over time to reduce redundant measurements in the computer system, comprising:

a server processing system identifying a group of performance parameters, of a multitude of performance parameters, for monitoring the computer system, including

the server processing system receiving performance indicators from a plurality of processes of the computer system,

the server processing system constructing a correlation matrix among the received performance indicators, wherein the correlation matrix is a table holding correlation values between said elements of the computer system,

the server processing system using the correlation matrix to construct a time-varying correlation graph indicating defined, changing correlations between a multitude of performance parameters of the computer system, and

the server processing system mining said defined correlations indicated in the correlation graph to identify monitoring logic for monitoring the elements of the computer system, including selecting a group of the performance parameters, from the multitude of performance parameters, for monitoring the elements of the computer system;

using the selected group of the performance parameters to monitor the elements of the computer system; and

the server processing system dynamically altering the group of performance parameters selected for monitoring the elements of the computer system as the time-varying correlation graph changes over time by selecting different groups of the performance parameters at different times for monitoring the elements of the computer system based on changes in the defined correlations indicated in the correlation graph to reduce redundant measurements in the computer system.

14. The method according to claim 13 , wherein:

the using the correlation graph further includes identifying a rate of change of the correlation graph using a defined criteria;

the using the selected group of the performance parameters includes using each of the selected groups of the performance parameters for monitoring the system for an associated monitoring time period, and changing the lengths of the monitoring time periods based on the rate of change of the correlation graph.

15. The method according to claim 14 , wherein:

the using the correlation graph further includes identifying a rate of change of the correlation graph using a defined criteria; and

the changing the group of performance parameters includes dynamically changing the group of performance parameters based on the rate of change of the correlation graph.

16. The method according to claim 13 , wherein the correlation graph identify the ones of the multitude of parameters that have a predefined correlation with others of the multitude of parameters.

17. A monitoring system for dynamic adaptive monitoring of a system of devices over time to reduce redundant measurements in the computer system, the monitoring system comprising:

one or more processing units configured for:

identifying a group of performance parameters, of a multitude of performance parameters, for monitoring the system of processes, including

receiving performance indicators from a plurality of devices of the system of devices,

constructing a correlation matrix among the received performance indicators, wherein the correlation matrix is a table holding correlation values between elements of the system of devices,

using the correlation matrix to construct a time-varying correlation graph indicating defined, changing correlations between a multitude of performance parameters of the system of devices, and

mining said defined correlations indicated in the correlation graph to identify monitoring logic for monitoring the elements of the system of devices, including selecting a group of the performance parameters as monitoring parameters, from the multitude of performance parameters, for monitoring the elements of the system of devices;

using the selected group of the performance parameters to monitor the elements of the system of devices; and

dynamically altering the group of performance parameters selected for monitoring the elements of the system of devices as the time-varying correlation graph changes over time by selecting different groups of the performance parameters at different times for monitoring the elements of the system of devices based on changes in the defined correlations indicated in the correlation graph to reduce redundant measurements in the system of devices.

18. The monitoring system according to claim 17 , wherein:

the using the correlation graph further includes

identifying a rate of change of the correlation graph using a defined criteria;

using each of the groups of the performance parameters to monitor the system of devices for an associated monitoring time period; and

changing the lengths of the monitoring time periods based on the rate of change of the correlation graph.

19. The monitoring system according to claim 17 , wherein the generating time varying correlation graphs includes:

during a first time period, monitoring a first plurality of said multitude of parameters of the system of devices; and

generating one of the correlation graphs from said first plurality of said parameters.

20. An article of manufacture comprising:

at least one tangible computer readable device having computer readable program code logic tangibly embodied therein for dynamic adaptive monitoring of a computer system over time to reduce redundant measurements in the computer system, said program code logic, when executing in one or more processing units, performing the following:

identifying a group of performance parameters, from a multitude of performance parameters, for monitoring the computer system, including

receiving performance indicators of a plurality of processes of the computer system,

constructing a correlation matrix among the received performance indicators, wherein the correlation matrix is a table holding correlation values between elements of the computer system,

using the correlation matrix to construct a time-varying correlation graph indicating defined, changing correlations between a multitude of performance parameters of the computer system, and

mining said defined correlations indicated in the correlation graph to identify monitoring logic for monitoring the elements of the computer system, including selecting to select a group of the performance parameters, from the multitude of performance parameters, as monitoring parameters for monitoring the elements of the computer system;

using the selected group of the performance parameters to monitor the elements of the computer system; and

dynamically altering the group of performance parameters selected for monitoring the elements of the computer system as the time-varying correlation graph changed over time by selecting different groups of the performance parameters at different times for monitoring the elements of the computer system based on changes in the defined correlations indicated in the correlation graph to reduce redundant measurements in the computer system.

21. The article of manufacture according to claim 20 , wherein:

the using the correlation graph further includes identifying a rate of change of the correlation graph using a defined criteria; and

the using the selected group of the performance parameters includes using each of the selected groups of the performance parameters for monitoring the system for an associated monitoring time period, and changing the lengths of the monitoring time periods based on the rate of change of the correlation graph;

the changing the group of performance parameters includes dynamically changing the group of performance parameters based on the rate of change of the correlation graph.

22. The article of manufacture according to claim 20 , wherein:

the generating time varying correlation graphs includes

during a first time period, monitoring a first plurality of said multitude of parameters of the system, and

generating one of the correlation graphs from said first plurality of said parameters; and

the using the correlation graphs includes

analyzing said one of the correlation graphs to select a second plurality of said parameters of the system, and

using said second plurality of said parameters to monitor the system during a second time period.