Workload Management With Resource Transfer Sequence Planned As A Function Of Ranking Of Resource Allocations

Abstract

A computer system includes computing resources and workload containers to which the resources can be allocated. A workload manager plans 1) a reallocation of the computing resources among said containers, and 2) a transfer sequence for units of the computing resources to implement said reallocation. The transfer sequence specifies an order in which the computing resource units requiring transfer are to reallocated among the containers.

What is claimed is: 
     
         1 . A computer system comprising:
 workloads;   computing resource units including hardware resource units;   resource containers to which said resource units can be assigned and within which said workloads execute; and   a workload manager configured to:   plan a reallocation of said computing resource units among said containers by making assignments of said computing resource units to respective containers, and   plan a transfer sequence specifying an order in which the computing resource units requiring transfer are to reallocated among the containers based on timing requirements for individual containers for receipt of the computing resources.   
     
     
         2 . The computer system of  claim 1 , wherein the transfer sequence is determined at least in part by consideration of a cost to accomplish individual transfers of resources and/or workloads between source and destination containers. 
     
     
         3 . The computer system of  claim 2 , wherein said cost is based at least in part on the time to accomplish said individual transfers. 
     
     
         4 . The computer system of  claim 2  wherein said cost is based at least in part on the computer resources required to accomplish said individual transfers. 
     
     
         5 . The computer system of  claim 1 , wherein the transfer sequence is determined at least in part by consideration of how quickly a computing resource unit can be transferred. 
     
     
         6 . The computer system of  claim 1 , wherein the transfer sequence is determined such that a first computing resource unit that can be transferred less quickly is transferred prior to the transfer of a second computing resource unit that can be transferred more quickly than the first computing resource unit but at a higher cost. 
     
     
         9 . The computer system of  claim 1  wherein said workload manager maintains an allocation list listing and indicating respective priorities for computer resource units planned to be allocated to that container. 
     
     
         12 . The computer system of  claim 1 , wherein said workload manager generates a benefit ranking of candidate source workloads for one of said units to be transferred, said benefit ranking being according to a benefit to each said source workload of retaining said unit for a longer time, said transfer sequence being planned in part as a function of said benefit ranking. 
     
     
         15 . The computer system of  claim 12 , wherein said reallocation is at least in part a weighted allocation and said benefit ranking is derived at least in part from the incremental weighted value or cost which may be associated with each said unit of resource which was requested by but not allocated to each said workload. 
     
     
         16 . A computer-implemented method comprising:
 executing workloads in respective containers by using hardware resources allocated to those containers;   preparing a reallocation plan for reallocating said hardware resources amongst said containers, said reallocation plan requiring transfer of at least some hardware resource units, said reallocation plan being based at least in part on workload priorities pre-assigned to said workloads prior to said preparing, said preparing involving explicitly assigning resource-unit priorities to resource units requiring transfer to implement said reallocation so as to differentiate least-needed resource units from most-needed resource units;   preparing a transfer sequence plan for implementing said reallocation plan, said transfer sequence plan specifying an order in which said resource units requiring transfer are to be transferred, said order being based at least in part on said resource-unit priorities; and   after preparing said transfer sequence plan, implementing said reallocation plan by implementing said transfer sequence plan.   
     
     
         17 . The method of  claim 16  wherein said resource-unit priorities are assigned according to the relative benefit of accomplishing or relative cost of delaying or both a transfer of each said unit according to its assignment for said reallocation. 
     
     
         18 . The method of  claim 16  wherein said resource-unit priorities are derived, at least in part, from said workload priorities. 
     
     
         19 . A system comprising non-transitory computer-readable media encoded with code configured to, when executed by a processor:
 plan a reallocation of computing resources to resource containers in which workloads are executing, said reallocation requiring transfer of at least some computing resource units, said planning being based at least in part on workload priorities pre-assigned to said workloads prior to said planning, the planning involving explicitly assigning resource-unit priorities to resource units requiring transfer to implement said reallocation so as to differentiate least-needed units from most-needed units;   plan a transfer sequence for implementing said reallocation, said transfer sequence having an order in which said resource units requiring transfer are to be transferred, said order being based at least in part on said resource-unit priorities; and   after said transfer sequence is planned, implement said reallocation by implementing said transfer sequence.   
     
     
         20 . The system of  claim 19 , wherein the transfer sequence is determined at least in part by consideration of how quickly a computing resource unit can be transferred.