Internal TFS server database maintenance

Brian posted these answers in a comment to a question on his January dogfood statistics post.  These may be of interest to folks out there, particularly those running larger-sized Team Foundation Server installations.

1) What is the current size of our dogfood database?

  • Version control: 928G
  • Workitem tracking: 121G
  • Warehouse: 17G
  • The rest is noise.

2) Backups/Restores – With large databases, are there any issues with backups and/or restores?  How often do we run backups?  How long do they take typically?

Full Databases backups (using LiteSpeed) – Every night

Transactional Logs backups (using LiteSpeed) – Every 15 minutes

For our biggest instance (DevDiv, total size a little bit over 1TB) – total full backup time for all databases is around 10-12 hours.

3) Integrity Checks – Do we run periodic “integrity checks” on the database and if so, how long do they take?

dbcc checkdb for all databases – every weekend (total time 3-5 hours for all databases)

4) Index Regeneration – Do we run periodic “index regeneration” on the database and if so, how long does it take?

Index defragmentation – every weekend (total time 2-3 hours for all databases)

Full index rebuilds have gotten very long and can’t be done online so we are looking at partitioning some of our largest tables to make that infrequent maintainence operation less disruptive.

tags: , ,

Comments (4)

  1. Buck Hodges 在日前整理了目前 內部 TFS 資料庫維運規劃方向, 對於 正在導入的夥伴來說, 這是很重要的 經驗分享, 特別需要強調. 我簡單以中文來描述: 目前的版本管控資料庫大約已經使用了

  2. Rob Caron on A Security Overview of Team Edition for Database Pros. Clara Oscura on MSF: track, workstreams,…

  3. Is this running on 32 or 64 Bit Sql Server?

  4. buckh says:

    It is 64-bit SQL Server 2005.