This is a double-post from my SQLBlogcasts.com blog site here
Allow me to share with you the results of a survey I did about how people are using virtualisation with SQL Server. This is a three part series of articles sharing the results.
In December 2010 I did a 10 question survey to see how people were using virtualisation with SQL Server and if they were changing their deployment and management processes as a result. 21 people responded from at least Europe, the US and Asia Pac. It’s not a massive number but it was large enough to see some trends and some differences of opinion as we’ll see.
This survey followed on from some thoughts Brent Ozar’s (web|twitter) presentation at SQL Bits gave me about virtualisation needing to change behaviours. He also helped promote my survey so thank you Mr Ozar.
Those who answered had a variety of interactions with SQL Server although the majority were SQL Server DBAs who also interact with the operating system, server hardware and storage. The dev users weren’t missed out representing approximately 40% of the participants. Most interesting was that only 9 out of 21 had any involvement with the hypervisor, arguably the most important component in a virtualisation stack.
What are people using virtualisation for?
I asked where in the application lifecycle people were using virtualisation and the graph below shows the answers best.
Dev and test are obviously biggest areas, no doubt because of the ability to spin-up, spin-down, clone and snapshot makes virtualisation a good platform for these areas. A lower adoption rate in production environments opens up a topic I cover later on: do people trust virtualisation yet for production, or are there features needed which virtualisation can’t deliver? Not everyone has a build, staging or DR environment so I wouldn’t suggest lower figures here are saying bad things about virtualisation.
Do people trust virtualisation used with SQL Server?
There’s a big perception still based upon nothing specific that virtualisation is still in its infancy and people generally don’t trust it. My survey couldn’t answer that question as it was for people who are already using virtualisation with SQL Server. However, what I did ask was how people who do use both together feel about it.
The pie chart below speaks for itself. Everyone is happy using virtualisation, they may not have had a choice even though they didn’t trust it, but what’s interesting is that a large share of people still don’t trust it for production environments.
What influence do people have on the hypervisor?
The results here were as I expected, the majority of people who answered had either full control over the hypervisor or could influence the sizing and configuration of their virtual servers. This is good as it means the sizing of servers still lies with the SQL Server teams, however do SQL Server teams know how to size their virtual servers is the next question?
Part two of this article is available here.