Using the Login Virtual Session Indexer (Login VSI) heavy workload, a 4U system consisting of the Avere FXT 4500 Edge filer and a NetApp FAS 2040 Core filer can handle nearly 1,500 VDI instances generating 17,000 IOPS with response times averaging eight milliseconds, achieving the lowest price per desktop with a benchmarked heavy workload of $11 per VDI IOP. With this unprecedented density of 4,250 IOPS/RackUnit, VDI deployments can be sized down to require minimal rackspace.
"Virtualization is an extremely valuable and flexible tool that brings tremendous efficiency to the data center, provided the storage infrastructure can handle the strains and challenges that situations, such as boot storms, produce to the system," said Ron Bianchini, Avere President and CEO. "Our goal has always been to provide the highest performance and capacity at the lowest cost and this VDI benchmarking test is a testament to the solutions we provide our customers. The numbers speak for themselves and we challenge anyone in the market to deliver the same results."
Method of Deriving Results
To measure the performance quality of 1,500 running instances of VDI, the Login VSI benchmark framework was used to drive real live I/O from within each of the VDI machines. The Login VSI benchmark measures the completion times of VDI desktop user tasks and monitors progress as the active VDI instance count grows. As soon as task-completion times exceed a threshold, the task-completion times of the active sessions are indexed and scored. With 1,480 active VDI instances running, the FXT 4500 Edge filer achieved a Dynamic VSImax score of 5050 while maintaining NFS Datastore response times below eight milliseconds.
Login VSI is a vendor independent benchmarking tool to objectively test and measure the performance and scalability of Virtual Desktop Infrastructures and Server Based Computing environments by simulating unique user workloads. Leading IT-analysts recognize and recommend Login VSI as the de-facto industry standard benchmarking tool for VDI and SBC.