Comments on: SQL Server Index Tuning Tip: Identify Overlaps

By: Clustered and Non Clustered Indexes « SSIS and Sql Server Journey

Clustered and Non Clustered Indexes « SSIS and Sql Server Journey — Mon, 06 Feb 2012 09:33:49 +0000

[...] tuning: http://www.brentozar.com/archive/2009/07/tuning-tip-identify-overlapping-indexes/ Share this:TwitterFacebookLike this:LikeBe the first to like this [...]

By: Jacob Buter

Jacob Buter — Fri, 28 Oct 2011 17:21:00 +0000

Brent,

Why don’t you include the field [RunID] in your new created index. Instead of going to the real data to find out if the [RunID] is the right one. This means less I/O’s. Or depends on the usage, you can add the [RunID] instead of [QuotaItemDriverID] in the indexed fields and add [QuotaItemDriverID] in the include fileds. This realy depends on the usage of the indexes.

Thanks,

Jacob

By: Brent Ozar

Brent Ozar — Thu, 18 Aug 2011 15:43:18 +0000

Reg84 – I talk about that in more detail in the video itself. Generally, no, I wouldn’t remove an index if it’s being used, but there’s always exceptions. For example, if it’s narrower than a wider index, and the wider one is being heavily used, it might make sense to remove the narrower one even if it’s being used. Every case is different, though, and it’s beyond what I could cover in a blog comment.

By: Reg84

Reg84 — Thu, 18 Aug 2011 03:51:36 +0000

Based on your video in sqlserverpedia on removing indexes, is it a good idea to remove a index with reads_per_write greater than 1. What is your suggestion?

Thanks.