We have recently completed a new performance study that focuses on the performance of Microsoft SQL Server 2019 on vSphere 7.0 U2. This blog has some of the highlights from that study. Please read the full document for additional testing details, key best practices for performance, and references to additional resources.
Microsoft SQL Server is one of the most widely used databases, serving up millions of query results every day for customers all over the world. Ensuring the good performance of SQL Server on VMware vSphere has been a constant focus here at VMware for many years. Previous studies and whitepapers with vSphere 5.5, 6.0, and 6.5 have been published that showed excellent performance of SQL Server on vSphere. This most recent paper uses vSphere 7.0 U2 and updates these previous papers.
This new performance study shows how vSphere 7.0 U2 running on a host configured with the latest 3rd Generation Intel Xeon Scalable Processors (known as “Ice Lake”) provides a robust platform on which to virtualize Microsoft SQL Server 2019.
Performance testing shows that vSphere 7 can handle tens of thousands of online transaction processing (OLTP) database operations per minute. This represents an increase in performance over previous generation servers.
In one set of “scale-up” tests, we look at how well a SQL Server 2019 instance running in a vSphere 7 VM performs as we increase the number of vCPUs. When scaling up, we want to see a near linear progression in orders per minute (as measured in the DVD Store 3.5 benchmark) as more vCPUs are added. Our test result in figure 1 shows that that going from 8 to 64 vCPUs raises the orders per minute as expected, coming up just below linear with over a 7x gain in throughput with an 8x gain in vCPUs. One additional set of tests was done with 76 vCPUs which matches the number of cores in the physical host.
For more information, read the paper Microsoft SQL Server 2019 on VMware vSphere 7.