Press "Enter" to skip to content

Category: Indexing

Don’t Over-Index

Published 2016-10-14 by Kevin Feasel

Kenneth Fisher shows the downside cost of non-clustered indexes:

I want to make a couple of final points. I realize 99 indexes is a lot. It’s to emphasize the differences. However they were also fairly small indexes and this is a single table where a normal database might easily have hundreds. So take these results as an example. They aren’t going to match real life but will hopefully show you how all of this can play out.

Indexes are awesome but you want to be smart about adding them. My personal rule of thumb, with no scientific evidence behind it, is 5 indexes or less and I’m pretty easy. 5-10 indexes and you’ll have to convince me. I’m going to be reviewing the existing indexes and see what I can get rid of, or maybe I can combine something. Past 10 indexes and it had best be for a query that’s running a 100+ times a minute or something for the CEO.

Read on for demo code and specific results.

Comments closed

Nonclustered Columnstore Indexes On Indexed Views

Published 2016-10-11 by Kevin Feasel

Niko Neugebauer notes that non-clustered columnstore indexes can now sit on top of indexed views, as of SQL Server 2016:

From the perspective of the disk access, this is where you will definitely win at least a couple of times with the amount of the disk access while processing the information, amount of memory that you will need to store and process (think hashing and sorting for the late materialisation phases), and you will pay less for the occupied storage.

Another noticeable thing was that the memory grants for the Indexed Views query was smaller compared to the query that was processing the original columnstore table FactOnlineSales.

Clustered indexes are currently not available as an option; we’ll see if that changes in the next version of SQL Server.

Comments closed

Sequentially Increasing Indexes

Published 2016-10-10 by Kevin Feasel

Joe Chang discusses benchmarking and looks at a particular scenario around maximizing insert performance:

The test environment here is a single socket Xeon E3 v3, quad-core, hyper-threading enabled. Turbo-boost is disabled for consistency. The software stack is Windows Server 2016 TP5, and SQL Server 2016 cu2 (build 2164). Some tests were conducted on a single socket Xeon E5 v4 with 10 cores, but most are on the E3 system. In the past, I used to maintain two-socket systems for investigating issues, but only up to the Core2 processor, which were not NUMA.

The test table has 8 fixed length not null columns, 4 bigint, 2 guids, 1 int, and a 3-byte date. This adds up to 70 bytes. With file and row pointer overhead, this works out to 100 rows per page at 100% fill-factor.

Both heap and clustered index organized tables were tested. The indexes tested were 1) single column key sequentially increasing and 2) two column key leading with a grouping value followed by a sequentially increasing value. The grouping value was chosen so that inserts go to many different pages.

The test was for a client to insert a single row per call. Note that the recommended practice is to consolidate multiple SQL statements into a single RPC, aka network roundtrip, and if appropriate, bracket multiple Insert, Update and Delete statements with a BEGIN and COMMIT TRAN. This test was contrived to determine the worst case insert scenario.

With that setup in mind, click through to learn his results.

Comments closed

Aborting Index Rebuilds

Published 2016-10-07 by Kevin Feasel

Arun Sirpal shows how to use the ABORT_AFTER_WAIT attribute on an index rebuild command:

Looking into the locking you will see that ONLINE operation uses (Sch-M) on the corresponding table as part of the process (actually takes Shared Table Lock (S) at the beginning of the operation, and a Schema Modification Lock (Sch-M at the end)).

So to be granted a SCH-M lock you can’t have any conflicting locks, so what happens when / if you have a process that is updating the table and you want to use the ONLINE rebuild? Yes you will be blocked. With 2014 onwards we can control what happens if we get into this situation and for this post I am going to abort the other query causing me to wait.

Not sure I like the “Kick the other guy(s) off” part that much, but I can see uses.  It’s probably more likely to go the opposite route, cancelling the rebuild if the server’s too hot.

Comments closed

Using Statistics For Index Design

Published 2016-10-07 by Kevin Feasel

Kendra Little argues that you should not use automatically created statistics as a guide for index creation:

We’ve talked a lot so far about how much statistics and indexes are related. This is why it seems like statistics might be useful for designing indexes!

But here’s the thing — SQL Server doesn’t track and report on how many times a statistic was used during optimization.

This is an interesting discussion.

Comments closed

Indexing Foreign Keys

Published 2016-09-26 by Kevin Feasel

Kendra Little answers a question about indexing foreign key constraints:

We recently had a SQL Server performance assessment. It remarked on two things that made me think:

1 ) tables with foreign keys but missing supporting index
2 ) tables with indexes that are not used

But in our case the remark in Case 2 was on a index that supports a foreign key!

Curtain down!

Now to my question, in which cases should you as a rule create indexes to support a foreign key?

I think Kendra elaborates the pros and cons well.  I’d lean against automatically creating indexes because I’ve worked with scenarios in which you don’t drive from the Parent to the Child; instead, you drive from Child to Parent or Something to Child to Parent and those indexes go unused.

Comments closed

Heap Pages Can Be Linked

Published 2016-09-23 by Kevin Feasel

Paul Randal gives a very special case in which heap pages can be (temporarily) linked together:

Now the pages are linked together!

Note that this is an OFFLINE rebuild, which is the default. What happened is that the offline ALTER TABLE … REBUILD operation uses the part of the underlying functionality for an offline ALTER INDEX … REBUILD operation that builds the leaf level of the index. As that functionality builds a doubly-linked list of pages, the newly rebuilt heap initially has a doubly-linked list of pages! This doesn’t happen for an ONLINE rebuild of the heap, which uses a totally different mechanism.

Although the pages appear doubly-linked, that’s just an artifact of the mechanism used to build the new heap – the linkages aren’t used or maintained.

In other words, you get zero benefit from this.  Click through to see Paul’s test script.

Comments closed

SARGable Predicates

Published 2016-09-15 by Kevin Feasel

Gail Shaw discusses what makes a particular predicate SARGable:

Any1 function on a column will prevent an index seek from happening, even if the function would not change the column’s value or the way the operator is applied, as seen in the above case. Zero added to an integer doesn’t change the value of the column, but is still sufficient to prevent an index seek operation from happening.

While I haven’t yet found any production code where the predicate is of the form ‘Column + 0’ = @Value’, I have seen many cases where there are less obvious cases of functions on columns that do nothing other than to prevent index seeks.

UPPER(Column) = UPPER(@Variable) in a case-insensitive database is one of them, RTRIM(COLUMN) = @Variable is another. SQL ignores trailing spaces when comparing strings.

This is a straightforward concept with significant performance implications.

Comments closed

Non-Clustered Index Design

Published 2016-09-13 by Kevin Feasel

Derik Hammer has an article on non-clustered index design:

In general, non-clustered indexes are a positive force in your database. Indexes are a form of data duplication, however, which costs disk space. In addition, non-clustered indexes need to be maintained. This can increase the number of writes which occur during INSERT or UPDATE operations and increase the number of index rebuilds or reorganizations that need to be performed.

Create non-clustered indexes to support all of your queries but be careful not to create duplicates and regularly purge indexes which are no longer necessary.

Worth reading the whole thing.

Comments closed