Chargeback reduction planning for information technology management

Abstract

Minimizing cost chargeback in an information technology (IT) computing environment including multiple resources. One implementation involves determining time-based usage patterns and allocation statistics for a plurality of resources and associated resource workloads. Using a regression function for determining a correlation of response time with resource usages and outstanding input/output instructions for the plurality of resources. Based on the time-based usage patterns, allocation statistics and the correlation, deriving an interpolation using positive and negative integrals to minimize a difference between allocated resource values and average allocation values. Determining service level objectives (SLOs) and resource allocation for minimizing cost chargeback for the resource workloads based on the derived interpolation.

1 . A method of minimizing cost chargeback in an information technology computing environment including multiple resources, comprising:
 determining time-based usage patterns and allocation statistics for a plurality of resources and associated resource workloads;   using a regression function for determining a correlation of response time with resource usages and outstanding input/output instructions for the plurality of resources;   based on the time-based usage patterns, allocation statistics and the correlation, deriving an interpolation using positive and negative integrals to minimize a difference between allocated resource values and average allocation values; and   determining service level objectives (SLOs) and resource allocation for minimizing cost chargeback for the resource workloads based on the derived interpolation.   
     
     
         2 . The method of  claim 1 , further comprising providing a recommendation for resource allocation that minimizes chargeback costs. 
     
     
         3 . The method of  claim 1 , wherein the interpolation comprises an objective function based on an integral area of the resource workload, cost rate for each device, and operation zone of each resource workload. 
     
     
         4 . The method of  claim 3 , wherein the positive integrals comprise an area of a curve above allocated resource values, and the negative integrals comprise an area of a curve between average allocated values and allocated resource values. 
     
     
         5 . The method of  claim 3 , wherein the objective function comprises: 
       
         
           
             
               
                 
                   
                     
                       Max 
                        
                       
                         
                           ∑ 
                           
                             j 
                             = 
                             0 
                           
                           devices 
                         
                          
                         
                           
                             ∑ 
                             
                               i 
                               = 
                               0 
                             
                             workloads 
                           
                            
                           
                             
                               A 
                               ij 
                               + 
                             
                             × 
                             
                               C 
                               j 
                             
                             × 
                             
                               SLO 
                               i 
                             
                           
                         
                       
                     
                     - 
                     
                       
                         ∑ 
                         
                           j 
                           = 
                           0 
                         
                         devices 
                       
                        
                       
                         
                           ∑ 
                           
                             i 
                             = 
                             0 
                           
                           workloads 
                         
                          
                         
                           
                             A 
                             ij 
                             - 
                           
                           × 
                           
                             C 
                             j 
                           
                           × 
                           
                             SLO 
                             i 
                           
                         
                       
                     
                   
                 
                 
                   
                     ( 
                     2 
                     ) 
                   
                 
               
             
           
         
         wherein 
         A ij =integral area of resource workload i at device j, where (A+): area of curve above allocated resource value, and (A−): area of curve in between average allocated value and allocated resource value, C j =Cost rate for device j, and SLO i =Operation zone of resource workload i. 
       
     
     
         6 . The method of  claim 5 , wherein constraints for the objective function comprise latency of service level objectives for each resource workload, latency based on outstanding input/output operations and load of outstanding input/output operations. 
     
     
         7 . The method of  claim 3 , wherein the objective function uses non-linear optimization that interpolates impact of resource allocation change on application latency in relation to SLO. 
     
     
         8 . The method of  claim 1 , wherein the regression function quantifies resource workload latency as a function of number of outstanding input/output operations. 
     
     
         9 . The method of  claim 1 , wherein the SLOs use randomized bin packing 
     
     
         10 . A system comprising:
 an evaluation module that evaluate time-based resource usage patterns and allocation statistics for a plurality of resources and associated resource workloads; and   a chargeback optimization module that determines a correlation of response time with resource usages and outstanding input/output instructions for the plurality of resources, and based on the time-based resource usage patterns, allocation statistics and the correlation, derives an interpolation using positive and negative integrals to minimize a difference between allocated resource values and average allocation values, and determines service level objectives (SLOs) and resource allocation for minimizing cost chargeback for the resource workloads based on the derived interpolation.   
     
     
         11 . The system of  claim 10 , further comprising an enterprise network coupled to the system. 
     
     
         12 . The system of  claim 10 , wherein the plurality of resources comprise:
 storage devices;   latency allocation;   at least one server device;   a plurality of switches; and   a plurality of applications executed by the at least one server device.   
     
     
         13 . The system of  claim 10 , wherein a final cost reduction recommendation is provided by the chargeback optimization module, wherein the final cost reduction recommendation comprises a time varying SLO. 
     
     
         14 . The system of  claim 13 , wherein the final cost reduction recommendation comprises one of a change of a current resource allocation value and a new resource allocation. 
     
     
         15 . The system of  claim 13 , wherein the final cost reduction recommendation defines resource allocation that minimizes chargeback costs. 
     
     
         16 . A computer program product for minimizing chargeback costs comprising a non-transitory computer usable medium including a computer readable program, wherein the computer readable program when executed on a computer causes the computer to:
 determine time-based usage patterns and allocation statistics for a plurality of resources and associated resource workloads;   use a regression function for determining a correlation of response time with resource usages and outstanding input/output instructions for the plurality of resources;   based on the time-based usage patterns, allocation statistics and the correlation, derive an interpolation using positive and negative integrals to minimize a difference between allocated resource values and average allocation values; and   determine service level objectives (SLOs) and resource allocation for minimizing cost chargeback for the resource workloads based on the derived interpolation.   
     
     
         17 . The computer program product of  claim 16 , wherein the computer readable program when executed on the computer further causes the computer to: provide a recommendation for resource allocation that minimizes chargeback costs. 
     
     
         18 . The computer program product of  claim 16 , wherein the interpolation comprises an objective function based on an integral area of the resource workload, cost rate for each device, and operation zone of each resource workload. 
     
     
         19 . The computer program product of  claim 18 , wherein the positive integrals comprise an area of a curve above allocated resource values, and the negative integrals comprise an area of a curve between average allocated values and allocated resource values. 
     
     
         20 . The computer program product of  claim 16 , wherein the objective function uses non-linear optimization that interpolates impact of resource allocation change on application latency in relation to SLOs.