Grid network management via automatic trend analysis of a service level agreement

1. A method for managing a grid network, comprising:

automatically performing trend analysis of a job type repeatedly processed by said grid network to anticipate at least one future load condition on said grid network associated with said job type, wherein said job type is associated with a service level agreement (SLA), wherein said performing of said trend analysis comprises:

processing a client defined test job, associated with said job type, in said grid network, wherein said test job is used for controlled testing of a performance of said grid network; and

measuring an external performance metric of said grid network that is associated with said processing of said test job;

automatically measuring at least one internal performance metric of said grid network to monitor current grid network status; and

automatically predicting future satisfaction of said SLA by comparing said at least one future load condition that is anticipated and said at least one internal performance metric.

2. The method of claim 1 , further comprising:

determining an action to be automatically performed by said grid network based on whether said SLA is satisfied under said at least one future load condition.

3. The method of claim 2 , further comprising:

performing said action.

4. The method of claim 2 , wherein said determining of said action comprises:

determining said action, wherein said action comprises a redistribution of data associated with said test job throughout said grid network.

5. The method of claim 2 , wherein said determining of said action comprises:

determining said action, wherein said action comprises re-allocating resources in said grid network assigned to jobs associated with said SLA.

6. The method of claim 1 , wherein said measuring of said external performance metric comprises:

measuring a return time for executing said test job.

7. The method of claim 1 , wherein said performing of said trend analysis comprises:

processing a real job, submitted by a client, by said grid network; and

measuring said external performance metric of said grid network that is associated with said processing of said test job.

8. The method of claim 1 , wherein said internal performance metric is associated with a node and is taken from a group consisting of: throughput, memory capacity, and resource utilization.

9. A method for managing a grid network, comprising:

automatically processing a plurality of jobs that is associated with a job type in said grid network, wherein said job type is associated with a service level agreement (SLA), wherein said processing of said plurality of jobs comprises:

processing a client defined test job, associated with said job type, by said grid network, wherein said test job is used for testing performance of said grid network; and

measuring an external performance metric of said grid network that is associated with said processing of said test job;

automatically determining a performance history of said grid network in processing said plurality of jobs;

automatically performing trend analysis to anticipate at least one future load condition for a future load on said grid network associated with said job type;

automatically measuring at least one internal performance metric of said grid network to monitor current GRID network status; and

automatically predicting future satisfaction of said SLA by comparing said at least one future load condition that is anticipated and said at least one internal performance metric.

10. The method of claim 9 , wherein said processing of said plurality of jobs comprises:

processing a real job, submitted by a client, by said grid network; and

measuring said external performance metric of said grid network that is associated with said processing of said test job.

11. The method of claim 9 , further comprising:

determining an action to be automatically performed by said grid network based on whether said SLA is satisfied under said future load.

12. The method of claim 11 , further comprising:

performing said action.

13. A system for managing a grid network, comprising:

a processor for executing instructions; and

a computer-readable storage medium with the instructions that the processor executes stored on the computer-readable storage medium, wherein the processor executes the instructions that implement:

a service level agreement (SLA) manager for performing trend analysis of a job type that is repeatedly processed by said grid network to anticipate at least one future load on said grid network, wherein said SLA manager submits a client defined test job of said job type for processing in said grid network, wherein said test job is used for controlled testing of a performance of said grid network and measures an external performance metric of said grid network that is associated with processing of said test job;

at least one resource manager for measuring at least one internal performance metric of a node of said grid network to monitor current grid network status; and

a predictive behavior modeling module for predicting satisfaction of a Service Level Agreement (SLA) by comparing said future load that is anticipated and said at least one internal performance metric.

14. The system of claim 13 , further comprising:

an action analyzer for determining an action, to be automatically performed by said grid network based on whether said SLA is satisfied under said future load.

15. The system of claim 13 , wherein said SLA manager monitors processing of a real job, submitted by a client, in said grid network, wherein said job type comprises said real job, and wherein said SLA manager measures an external performance metric of said grid network that is associated with processing of said real job.

16. The system of claim 13 , wherein said SLA manager further comprises:

a profile builder for determining predictable service levels under said at least one future load.

17. The system of claim 13 , further comprising:

a metadata repository for storing results from said trend analysis and said at least one internal performance metric.