Lab Validation: EMC Symmetrix VMAX and Microsoft Server Virtualization Scalability
When the implementation was configured with a maximum implementation, 1024 virtual machines were distributed across 16 high performance servers. Each server was configured with two quad-core high speed Intel processors and 96 GB of RAM. The servers were connected to EMC VMAX storage through a dual fabric 8Gb Brocade SAN. Multiple paths to the VMAX storage were configured from each server for added scalability and availability. The storage was carved into three thin pools comprised of a total of 702 drives in RAID-1 mirror sets. Thin devices were used to create 24 1 TB 16-way striped data LUNs. The 24 LUNs were thin provisioned and presented to the servers as CSVs. On the volumes, 1,024 1 TB Hyper-V R2 virtual hard drives were used to host virtual machine operating system images.
FE Bandwidth (Gbps)
Usable Capacity (TB)
The configuration used to deploy 1,024 virtual machines was analyzed by ESG Lab with a goal of estimating the server and storage horsepower needed to support smaller virtual server deployments. The results, which are summarized in Table 1, can be used to plan the resources required to get started with a server virtualization solution that can scale to meet the needs of the business. For example, the second column shows how a pair of servers attached to two VMAX engines can be used to support up to 128 server-class virtual machines. Columns to the right illustrate how servers, VMAX engines, and drives can be added to support up to 1,024 virtual machines with a consolidated Enterprise-class infrastructure.
Table 1. Scaling to 1,024 Virtual Machines in One Microsoft Hyper-V R2 Cluster with EMC VMAX
Why This Matters
Managing server sprawl is a challenge faced by many mid- and large-scale IT organizations. Under-utilized servers take up space, waste resources, and are more likely to be improperly aligned with the workloads they support. Server virtualization can be used to combat server sprawl as it consolidates the number of servers in the data center.
Despite the broad success of server virtualization, a new set of scalability challenges has emerged. As a result, the consolidation ratios of virtual machines per physical server (and the savings that can be achieved) have remained relatively low. If the scalability challenges can be addressed, consolidation ratios—and savings—are bound to rise as well.
ESG Lab has confirmed that a cluster of Hyper-V R2 multi-core servers connected to an EMC VMAX with multiple storage engines can be used to meet the scalability challenges associated with wide-scale virtual server consolidation.
© 2010, Enterprise Strategy Group, Inc. All Rights Reserved.