Абстрактный

K-Tier Computation of Parallel Workload in Cloud

P.Karthikeyan, R.Sudhakar

Cloud computing model attracts many complex applications that run on data centers. Complex applications often require parallel processing capabilities. By using virtualization technologies the computing capacity of each node is divided into K-Virtual Machines (KVMs). Attaining KVMs are based on the available resources of physical machine. Collocation of virtual machines on each node will run multiple jobs simultaneously. Running parallel jobs on each node would increase resource utilization on each physical machine. Job management is the key role in cloud computing systems, parallel job scheduling problems are main which relate to the efficiency of the whole cloud computing. Here, the Virtual Machines (VMs) are of two types, foreground VM and background VMs. Foreground VM will have high priority whereas background VMs has low priority. By Modifying Conservative Migration Supported Backfilling algorithm, we can run parallel jobs on each node with K-VMs to improve utilization of available resource.