Dell published a very short but interesting paper about performances of Microsoft SQL Server 2005 in a VMware ESX Server 2.5.2 infrastructure:
To investigate the performance of SQL Server 2005 in a server consolidation environment, in May 2006 the Scalable Enterprise Technology Center compared the performance of 15 VMs running SQL Server 2000 with 15 VMs running SQL Server 2005 on a Dell PowerEdge 2850 server with two dual-core Intel Xeon processors at 2.8 GHz and 8 GB of RAM. Each VM had the same configuration except for the version of SQL Server, with 512 MB of RAM, a 10 GB hard disk, and a vmxnet virtual Gigabit Ethernet4 NIC.
ESX Server 2.5.2 was installed on the PowerEdge 2850 server, and Windows Server 2003 Enterprise Edition with SP1 was the guest OS on all the VMs. The number of VMs running on the server was set at 15, which was based on findings in the Dell white paper VMware ESX Server Performance Gains on Dell PowerEdge 2850 Dual Core Servers…
One of the most critical service to move in a virtual environment is database depending on its high I/O workloads, and at today I still see so many customers refusing to migrate.
Benchmarks like this one should help clarifying when is sensible use a virtual database and when not.
Unfortunately this study just compared performances between virtualized SQL Server 2000 and 2005, and not performances between a physical SQL Server 2005 and a virtual one.
Read the whole paper at source.
Thanks to VMTN Blog for the news.