Everything you want to know about Visual Studio ALM and Farming
Brian Harry is a Microsoft Technical Fellow working as the Product Unit Manager for Team Foundation Server. Learn more about Brian.
More videos »
It’s been about 6 months since I last posted detailed statistics on the DevDiv TFS dogfood server. Overall it’s been a good 6 months. Early this year we got the bulk of our TFS 2010 branching/merging/scaling changes deployed to the server and they’ve made a a big difference in the scalability challenges we were having last fall. The server’s running pretty well these days.
Later this month we’ll be setting up a second DevDiv dogfood server that will mirror chunks of the “main” server and will be used by the Visual Studio Team System team. This new separate server will exist because we need a way that we can use pretty recent builds without disrupting the work environment for the broader division with frequent updates. I’ll write some more about this in the next few weeks as the rollout progresses.
Here are some things that stand out to me when I look at the numbers:
1) Checkins are way up. My guess is that’s due to the fact that we are in an intense bug fixing period right now and we are probably seeing a whole lot of very small checkins.
2) File downloads are way down. Every since we put in the 3 node proxy to handle the download volume, the download numbers just haven’t been relevant any more. I don’t have any good way to collect statistics from the proxies (they aren’t instrumented as well as the main server is). I think this may be the last report that I include download numbers in because they just aren’t very relevant.
3) Files and Work item versions continue their astonishingly steady climb.
4) Workspaces are way up. In the fall and early this year we were aggressively deleting old unused workspaces in preparation for a series of server upgrades we were doing. We’ve pretty much let them go the last several months so they’ve been piling up.
5) I’ve stopped counting files & folders separately. For one thing the number has gotten huge. Secondly the TFS 2010 schema is less optimized to counting the difference. It basically involves a > 600 million row table scan on our data base and that’s way too long.
Commands (last 7 days)