Blog Post

How Many Drive Letters Do You Use for SQL Server?

,

I was teaching a private class recently that had both DBA's and network types in the room, one of the items that came up for discussion was about the 'correct number of drive letters' that should be used. The DBA's had set up separate drives for:

  • Master & MSDB
  • TempDB
  • Data
  • Logs
  • Backup

My standard setup is as follows:

  • Z drive - data
  • Y drive - logs
  • X drive - backup

So which is right? Would it be wrong to say it depends?

The thinking of the DBA team was that by splitting things out in advance, if they needed more space/performance they go could back to the network team and just have them re-provision whichever drive was causing a problem without having to make any configuration changes. That's not bad thinking, but it feels a little overdone in this case. It's not that drive letters cost money, but there are times when you can run short of them if you have a lot of data and you're clustering (quorum drive). Planning for the future isn't a bad approach, and it's a very Zen decision to know when you're preparing and when you're just adding complexity. I'm sure that many would disagree with my minimalist approach, but I've always started with the simplest configuration and if I need to move something (TempDB or just a really active DB), then I make the change when it's needed.

The more important point is to settle on a standard configuration and keep using it. I'm a firm believer in using patterns/muscle memory to keep us safe when we get forced into making changes when we're tired/under stress or otherwise distracted. It also simplifies internal communication, I can just tell the backup team to 'backup X on every SQL Server and that's all we need as far as data backup'...DR obviously requires a wider look.

One of the other points raised during this discussion (which was pleasant by the way) was about potentially running out of space. I use autogrow for data and logs, and I think most people do, but it does leave the possibility that the drive can be filled to capacity. Active monitoring should catch this, but there's still the chance for something to go wrong. I've long adopted the practice of leaving a placeholder file on each drive that I can delete on a really rainy day, has saved me pain a couple times.

 

Rate

You rated this post out of 5. Change rating

Share

Share

Rate

You rated this post out of 5. Change rating