Optimizing Workload Performance in Virtualized Environment

Virtualization has offered the ability to execute diverse applications (ERP, Exchange,Database etc..) on parallel VM’s in a shared hardware platform. This in turn has imposed challenges w.r.t. efficient resource scheduling/allocation to cater CPU and Network intensive workloads with several traits- random vs sequential, large vs small I/O request size, read vs. write ratio, and degree of parallelism. In this paper, we will explore the factors impacting the performance of virtual workloads in terms of Throughput, IOPs, Latency and enable below Automatic tunables to achieve application-optimized performance-: 1) Workload Profiling- Characterizing the workloads based on the data access, block size etc.2) QoS- Enabling Quality of service at the VM level to control resource multiplexing and resource scheduling particularly during Spikes 3) Distributed Data Tiering- Intelligent data placement across SSD-HDD tiers with negligible impact to real-time performance.