Method and apparatus for performance and policy analysis in distributed computing systems
AssigneeInternational Business Machines Corporation
InventorsMurthy V. Devarakonda, Nithya Rajamani, Mudhakar Srivatsa
Life of this patent
Estimating the term…
Abstract
One embodiment of the present method and apparatus for performance and policy analysis in distributed computing systems includes representing a distributed computing system as a state transition model. A queuing network is then superimposed upon the state transition model, and the effects of one or more policies on the distributed computing system performance are identified in accordance with a solution to the queuing network.
Claims18
A method for analyzing one or more policies applicable to a plurality of computing sites that process respective workloads in a distributed computing system, said method comprising:
- representing said distributed computing system as a state transition model;
- wherein said representing comprises:
- constructing said state transition model in accordance with at least one model representing characteristics of said plurality of computing sites, where said at least one model includes at least one of:
- a site model, a workload model, a workload state model, a site state model, a policy model, an event model or a cost model;
- wherein said representing further comprises:
- modeling, at a computing site, an availability of associated resources as a state;
- modeling, at said computing site, changes to an associated workload and changes to said availability of associated resources as state transitions;
- and
- modeling policies that dictate a manner in which said computing site shares resources with a second computing site as constraints on said state and on said state transitions;
- superimposing a queuing network model on said state transition model, wherein the superimposing said queuing network model on said state transition model comprises:
- identifying valid states and state transitions for the distributed computing system;
- and
- annotating the valid state transitions in said state transition model with probabilities at which the valid state transitions occur and probabilities associated with the policies;
- and
- identifying the effects of said one or more policies on said distributed computing system performance in accordance with a solution to said queuing network model.
A non-transitory computer readable medium containing an executable program for analyzing one or more policies applicable to a plurality of computing sites that process respective workloads in a distributed computing system, where the program performs the steps of:
- representing said distributed computing system as a state transition model;
- wherein said representing comprises:
- constructing said state transition model in accordance with at least one model representing characteristics of said plurality of computing sites, where said at least one model includes at least one of:
- a site model, a workload model, a workload state model, a site state model, a policy model, an event model or a cost model;
- wherein said representing further comprises:
- modeling, at a computing site, an availability of associated resources as a state;
- modeling, at said computing site, changes to an associated workload and changes to said availability of associated resources as state transitions;
- and
- modeling policies that dictate a manner in which said computing site shares resources with a second computing site as constraints on said state and on said state transitions;
- superimposing a queuing network model on said state transition model, wherein the superimposing said queuing network model on said state transition model comprises:
- identifying valid states and state transitions for the distributed computing system;
- and
- annotating the valid state transitions in said state transition model with probabilities at which the valid state transitions occur and probabilities associated with the policies;
- and
- identifying the effects of said one or more policies on said distributed computing system performance in accordance with a solution to said queuing network model.
Apparatus for analyzing one or more policies applicable to a plurality of computing sites that process respective workloads in a distributed computing system, said apparatus comprising:
- means for representing said distributed computing system as a state transition model;
- wherein the representing comprises:
- constructing said state transition model in accordance with at least one model representing characteristics of said plurality of computing sites, where said at least one model includes at least one of:
- a site model, a workload model, a workload state model, a site state model, a policy model, an event model or a cost model;
- wherein the representing further comprises:
- modeling, at a computing site, an availability of associated resources as a state;
- modeling, at said computing site, changes to an associated workload and changes to said availability of associated resources as state transitions;
- and
- modeling policies that dictate a manner in which said computing site shares resources with a second computing site as constraints on said state and on said state transitions;
- means for superimposing a queuing network model on said state transition model, wherein the superimposing said queuing network model on said state transition model comprises:
- identifying valid states and state transitions for the distributed computing system;
- and
- annotating the valid state transitions in said state transition model with probabilities at which the valid state transitions occur and probabilities associated with the policies;
- and
- means for identifying the effects of said one or more policies on said distributed computing system performance in accordance with a solution to said queuing network model.
Dependencies are inferred from the claim text.