“Hello, and welcome to our DBAs Anonymous meeting. Today we have a new visitor. Would you like to introduce yourself”
“Hello, my name is John. I’m a DBA and I don’t know how to cluster servers.”
“GASP!!! Get out!!!”
OK, so maybe it’s not quite like that. But sometimes that’s how I picture it in my mind. Until a few weeks ago I have never installed Microsoft SQL on a failover cluster. In my defense I’ve never worked in an environment that used clustering. I always felt that I was missing a big part of my SQL education. So I finally decided to do something about it.
I’ve been running Hyper-V on my laptop for a while now and I’ve already set up my virtual network. I installed and configured a domain controller running Windows 2008 R2 and two other servers that run SQL 2008 R2 and SharePoint 2010, also a few workstations that I use mostly for demos and small projects. Until recently I didn’t know how to create virtual disks that I could use for a test cluster. I did some research and did know that I could download the iSQSL Software Target, which I did on another server attached to my virtual network.
Recently I found two blogs that showed exactly how to configure the disks and set up the cluster. Prashant Kumar shows how to do this in Windows 2008 R2 on his blog SQLActions. And Ayman El-Ghazali does the same with Windows 2012 Core on his blog The SQL Pro. Between these two resources I was able to configure my lab with a cluster. I’ve built and destroyed a few dozen over the last two weeks, along with different SQL installations. So far I’m having a blast! I know, I know, wait until it’s production. But still…
I’ve just got a few additional tips to anyone else who stumbles across this post and wants to try the same.
First, if you use the differencing disk feature of Hyper-V, make sure you install all OS features, upgrades and patches to your base disk before you create your new virtual machines. This will save you hours, depending on how up to date you want your servers to be. In my case I installed PowerShell because I know I’ll be using it everywhere. On my node servers I put in other roles and features, but those won’t necessarily go on all my future servers.
Second, take advantage of snapshots in Hyper-V, again it will save you time when you want to redo an action or just start from scratch. In my case I took snapshots when I created my virtual machines, when I attached the storage, when I created the cluster, and when I first installed SQL. Now, for example, when I want to reinstall SQL with different options, I can easily get back to a fresh environment where I know there won’t be any conflicts from past installs.
I can show my method step by step if anyone is interested, but there’s enough information in the two blogs I mentioned earlier to get you started. Give me a few more weeks and I hope to be able to discuss SQL clustering here as well.