Energy Efficiency in Cloud Data Centers: Energy Efficient VM Placement for Cloud Data Centers Doctoral Student : Chaima Ghribi Advisor : Djamal Zeghlache Co-Author : Makhlouf Hadji Wireless Networks and Multimedia Services Department CNRS UMR 5157-Samovar, Telecom SudParis Summary page 1 Objectives Proposed Algorithms Evaluation Conclusion ICWS 2011, Washington DC, USA. Implementation of Communities of Webfor Service Registries Energy Efficient VM Placement Cloud Data Centers Objectives Energy aware VM placement in cloud data centers. Propose optimal algorithms for VM allocation and migration to reduce power consumption in cloud data centers page 2 ICWS 2011,ou Washington direction servicesDC, USA. Implementation of Communities of Webfor Service Energy Efficient VM Placement CloudRegistries Data Centers <pied de page> Framework • Energy-aware VM placement o Responsible for the optimal energy aware VM placement in the data center. • Energy consomption estimator o Relies on energy estimation tools that use power models to infer power consumption of VMs or servers from resource usage • Cloud Iaas manager o OpenStack, OpenNebula, CloudStack o Control and manage cloud resources, handle clients requests, schedule and provisioning of VMs page 3 ICWS 2011,ou Washington direction servicesDC, USA. Implementation of Communities of Webfor Service Energy Efficient VM Placement CloudRegistries Data Centers <pied de page> Proposed algoritms Exact VM placement algorithm o selects where to deploy VMs Exact VM Migration algorithm o migrates VMs to achieve consolidation Adapted energy aware best fit algorithm o used for comparison purposes page 4 ICWS 2011,ou Washington direction servicesDC, USA. Implementation of Communities of Webfor Service Energy Efficient VM Placement CloudRegistries Data Centers <pied de page> VM placement algorithm Objective, conditions & constraints Objective o initial VM placement leading to minimum number of used servers (or containers) Mathematical Programming Formulation o modelled as a bin packing problem with a minimum power consumption objective page 5 ICWS 2011,ou Washington direction servicesDC, USA. Variable comment m Number of servers Pj,Max Server power consumption limit Pj, current Current power consumption ej Boolean = 1 if j hosts VM xij Boolean = 1 if VM I assigned to server j n Number of requested VMs Implementation of Communities of Webfor Service Energy Efficient VM Placement CloudRegistries Data Centers <pied de page> VM placement algorithm Model variables 6/20 ICWS 2011, Washington DC, USA. Implementation of Communities of Web Service Registries VM Migration algorithm Objective o Optimize data center power consumption using dynamic VM consolidation Mathematical Programming Formulation o Based on linear integer programming formulation • Zijk = 1 if VM k migrated from server i to j • yi = 1 if server i idle and = 0 if at least one VM is active • m’ = number of non idle servers m’< m • P’k = power cost when migrating VM k • qi is the total number of VMs hosted on server i and candidate for migration into destination servers, especially server j in equation page 7 ICWS 2011,ou Washington direction servicesDC, USA. Implementation of Communities of Webfor Service Energy Efficient VM Placement CloudRegistries Data Centers <pied de page> VM Migration algorithm Maximize number of empty servers to shut them down by migrating VM to achieve consolidation if a VMk is migrated from a server i (source) to a server j (destination), it can not be migrated to any other server l (l j). Ensuing migrations forbidden Destination VM power budget limit has to be respected Non idle servers candidate for migration have to be entirely emptied Equivalent total number of empty servers Do not migrate a VM whose job is about to end…. page 8 ICWS 2011,ou Washington direction servicesDC, USA. Implementation of Communities of Webfor Service Energy Efficient VM Placement CloudRegistries Data Centers <pied de page> VM Migration algorithm A server candidate to a migration should not migrate its own VMs page 9 ICWS 2011,ou Washington direction servicesDC, USA. A VM can not be migrated to many servers at the same time Implementation of Communities of Webfor Service Energy Efficient VM Placement CloudRegistries Data Centers <pied de page> Adapted energy aware best fit algorithm Adaptation of the Best-Fit heuristic which consists of : Sorting items (VMs) in a decreasing sequence of their power consumption. Place all the sorted VMs by considering the first item (VM) in a server with a minimum remaining power budget page 10 ICWS 2011,ou Washington direction servicesDC, USA. Implementation of Communities of Webfor Service Energy Efficient VM Placement CloudRegistries Data Centers <pied de page> Evaluation Proposed algorithms evaluated using the linear solver CPLEX Estimate expected percentage of energy or power consumption savings when combining the exact allocation and migration algorithms page 11 ICWS 2011,ou Washington direction servicesDC, USA. Implementation of Communities of Webfor Service Registries Energy Efficient VM Placement Cloud Data Centers <pied de page> Evaluation page 12 Comparison between Exact Placement and Best Fit algorithms ICWS 2011,ou Washington direction servicesDC, USA. Implementation of Communities of Webfor Service Registries Energy Efficient VM Placement Cloud Data Centers <pied de page> Evaluation Performance comparison of the exact placement algorithm with and without migration page 13 ICWS 2011,ou Washington direction servicesDC, USA. Implementation of Communities of Webfor Service Energy Efficient VM Placement CloudRegistries Data Centers <pied de page> Evaluation page 14 Convergence time of the Exact Placement Algorithm ICWS 2011,ou Washington direction servicesDC, USA. Implementation of Communities of Webfor Service Energy Efficient VM Placement CloudRegistries Data Centers <pied de page> Evaluation Convergence time of the Exact Migration Algorithm (m’=5) Convergence time of the Exact Migration Algorithm (m’=10) page 15 ICWS 2011,ou Washington direction servicesDC, USA. Convergence time of the Exact Migration Algorithm (m’=20) Implementation of Communities of Webfor Service Energy Efficient VM Placement CloudRegistries Data Centers <pied de page> Evaluation page 16 Percentage of gained energy when migration is used ICWS 2011,ou Washington direction servicesDC, USA. Implementation of Communities of Webfor Service Energy Efficient VM Placement CloudRegistries Data Centers <pied de page> Conclusion Room for additional energy savings in data centers through even more efficient algorithms – joint / one shot scheduling and placement with reduced need for consolidation Pursue the quest for more efficient algorithms Looking currently at scheduling and placement leading to minimum power or energy consumption using graph coloring techniques page 17 ICWS 2011,ou Washington direction servicesDC, USA. Implementation of Communities of Webfor Service Energy Efficient VM Placement CloudRegistries Data Centers <pied de page> Published Paper Chaima Ghribi, Makhlouf Hadji, Djamal Zeghlache, "Energy Efficient VM Scheduling for Cloud Data Centers: Exact Allocation and Migration Algorithms," ccgrid, pp.671-678, 2013 13th IEEE/ACM International Symposium on Cluster, Cloud, and Grid Computing, 2013
© Copyright 2026 Paperzz