Indexing

  • A request from a user:

    ==

    I second the request for more articles on indexing! As junior-ish DBA to a new company (and the only real DBA), I had to evaluate various performance bottlenecks fairly quickly. Turns out that indexing was a major issue, with the most important table having over 55 indexes, and the top 10 tables having over 20 indexes each. I really could not find any solid information on how to begin to attack this problem - each article I came across gave me only bits and pieces of the puzzle. Specifically, which metrics to look at when evaluating, tuning existing indexes, and how to choose which ones to kill. I was eventually able to come up with a game plan, but something comprehensive might help someone else in the future!

    ==

    I think there's room for a few articles here, tackling this from new angles.

  • I would like to write an article for this. I recently had to address a real world issue regarding latching and blocking that was caused by parallelism resulting from poor indexing on tables. Things that would be covered in this article:

    -Index scans vs seeking

    -MaxDOP/Threads/Parallelism

    -Wait Statistics

    -Graphical execution plans in SSMS

    -DMV's to find cached sprocs using specific indexes, amount of times a sproc executed

    -DMV's to find indexes on tables, how many times indexes are utilized in execution plans, and seeks/updates/scans per index

    The items listed above would be covered in respect to ID'ing parallelism, diagnosing it as the bottleneck, and remedying it with indexes.

    Thanks

    David Fundakowski

  • Love to see it. Focus down on your problem, not generally how other systems might work. IF you can mock your situation, and process for solving, that's a great learning piece.

  • I have sent a PM with the article.

Viewing 4 posts - 1 through 3 (of 3 total)

You must be logged in to reply to this topic. Login to reply