Press "Enter" to skip to content

Category: Statistics

Filtered Statistics and Table Performance

Published 2022-07-11 by Kevin Feasel

Guy Glantser provides a use case for filtered statistics:

Let’s say you have a very large table on a SQL Server 2012 Standard Edition instance. This means: old cardinality estimator and no partitioning. The table has a DATETIME column, which is ever-increasing, and it contains 5 years of data (it has to, due to regulations). The auto-update statistics kicks in only every 4 days, more or less, even when trace flag 2371 is enabled. The problem is that users usually query the table for the last day, and only rarely need to access older data. Since auto-update statistics uses a very small sample rate for very large tables, the result is not very accurate. The bottom line of all this is that most of the time you get a poor execution plan, because the optimizer estimates very few rows, while in fact there are many rows. What can you do?

I’m not sure I’ve ever used filtered statistics but it is good to know such a thing exists.

Comments closed

Trivial Plans and Stats Updates

Published 2022-03-08 by Kevin Feasel

Lonny Niederstadt lays out the harshness of reality:

OK.  SQL Server trivial plans for rowstore table INSERT. And related optimizer stats interaction.
TL;DR cached trivial plans for INSERT can be surprisingly stubborn. If a query matches to one, it won’t perform or queue a stats update even if the stats are stale.  If the stats have been updated and would otherwise warrant a per-index plan – but there is a matching cached trivial plan for a per-row plan… outta luck. Might hafta DBCC FREEPROCCACHE or add OPTION(RECOMPILE) hint to make sure a cached trivial plan doesn’t prevent a per-index update for an INSERT when you really want one.

Read on for a dive into the topic.

Comments closed

Querying Stats Data with a DMF

Published 2022-02-08 by Kevin Feasel

Grant Fritchey wants queryable data:

We’ve always been able to look at statistics with DBCC SHOW_STATISTICS. You can even tell SHOW_STATISTICS to only give you the properties, STAT_HEADER, or histogram, HISTOGRAM. However, it’s always come back in a format that you can’t easily consume in T-SQL. From SQL Server 2012 to everything else, you can simply query sys.dm_db_stats_properties to get that same header information, but in a consumable fashion.

Read on for a quick post showing a couple of things you can do with the DMF.

Comments closed

Stats Q&A

Published 2021-11-17 by Kevin Feasel

Erin Stellato has a two-parter on statistics in SQL Server. Part 1 deals with questions on stats creation:

Last week I presented a session, Demystifying Statistics in SQL Server, at the PASS Community Summit, and I had a lot of great questions; so many that I’m creating multiple posts to answer them. This first post is dedicated to questions specific to creating statistics in SQL Server.

Part 2 deals with stats updates:

Last week I presented a session, Demystifying Statistics in SQL Server, at the PASS Community Summit, and I had a lot of great questions; so many that I’m creating multiple posts to answer them. This second post is dedicated to questions specific to updating statistics in SQL Server. Of note…I have a couple previous posts which also include helpful information:

Click through for lots of questions and lots of good answers.

Comments closed

All about Synchronous Stats Updates

Published 2021-10-26 by Kevin Feasel

Paul Randal shares some thoughts about synchronous stats updates:

The SQL Server query optimizer makes use of statistics during query compilation to help determine the optimal query plan. By default, if the optimizer notices a statistic is out-of-date because of too many changes to a table, it will update the statistic immediately before query compilation can continue (only the statistics it needs, not all the statistics for the table).

Note that “too many” is non-specific because it varies by version and whether trace flag 2371 is enabled – see the AUTO_UPDATE_STATISTICS section of this page for details.

Read on to learn more, including the problems that synchronous stats updates can cause, what you can do to avoid them, and ways you can tell that synchronous stats updates are a problem in your environment.

Comments closed

Enabling Statistics Auto-Creation

Published 2021-10-25 by Kevin Feasel

Chad Callihan checks the stats:

When we query for data, we don’t always think about the magic that goes into efficiently returning results. One vital piece to this magic is statistics. Statistics in SQL Server are histograms that are used by the query optimizer to determine an optimal execution plan when executing a query. Let’s take a look at the different ways to check your statistics settings and make sure statistics are being automatically created.

Click through to see how.

Comments closed

Ignoring Updates to Some Statistics

Published 2021-06-16 by Kevin Feasel

Raul Gonzalez gives some tips on optimizing statistics updates:

For now, everything described might not be such a horrible thing, it’s clear that SQL Server will not take full advantage of the stats on the column [Body] if the queries we are running use wildcards (specially leading), but why so much fuss? Well, now it’s when things start making sense (or not).

Running stats maintenance on this kind of columns every night can become really expensive and this is what I’ve found more than once when using the Query Store to look for queries that have a high number of reads.

Read the whole thing.

Comments closed

Viewing Stats Used in Creating Execution Plans

Published 2021-03-04 by Kevin Feasel

Matthew McGiffen shows us how to find the statistics used when generating an execution plan:

Statistics are vital in allowing SQL Server to execute your queries in the most performant manner. Having a deep understanding of how the SQL Server Optimizer interacts with Statistics really helps when you are performance tuning

One thing that can be useful when looking at an execution plan is to understand what statistics objects the optimizer used to come up with the plan. In this post we look at how that can be achieved using the undocumented traceflag 8666 which can be used to save internal debugging informational into the plan XML – including details of the Statistics objects used. 

Click through for a couple of caveats about this, as well as a primer on how to see those precious statistics.

Comments closed

Number of Rows Automatically Sampled versus Table Size

Published 2021-02-26 by Kevin Feasel

Matthew McGiffen does the math:

I mentioned in my previous post about manually updating statistics that you can specify whether they’re updated using a full scan, or you can specify an amount of data to sample, either a percentage of the table size, or a fixed number of rows. You can also choose not to specify this, and SQL Server will decide for you whether to do a full scan, or to sample a certain amount of data.

I thought it would be interesting to look at what the sample sizes are that SQL will choose to use, depending on the amount of data in your table. 

Click through for the result of Matthew’s analysis.

Comments closed