Category: Query Tuning

Kendra Little digs into a tricky performance problem:

Most of the time in SQL Server, the MAX() function and a TOP(1) ORDER BY DESC will behave very similarly.

If you give them a rowstore index leading on the column in question, they’re generally smart enough to go to the correct end of the index, and — BOOP! — just pluck out the data you need without doing a big scan.

I got an email recently about a case when SQL Server was not smart enough to do this with MAX() — but it was doing just fine with a TOP(1) ORDER BY DESC combo.

The question was: what’s the problem with this MAX?

It took me a while to figure it out, but I finally got to the bottom of the case of the slow MAX.

Great story and very interesting sleuthing work on Kendra’s part.

Paul White continues his row goals series:

The optimizer assumes that people write a semi join (indirectly e.g. using EXISTS) with the expectation that the row being searched for will be found. An apply semi join row goal is set by the optimizer to help find that expected matching row quickly.

For anti join (expressed e.g. using NOT EXISTS) the optimizer’s assumption is that a matching row will not be found. An apply anti join row goal is not set by the optimizer, because it expects to have to check all rows to confirm there is no match.

If there does turn out to be a matching row, the apply anti join might take longer to locate this row than it would if a row goal had been used. Nevertheless, the anti join will still terminate its search as soon as the (unexpected) match is encountered.

Another very interesting part of the series and well worth the time to read.

Bert Wagner compares a couple methods for writing a query:

Suddenly those key-lookups become too expensive for SQL Server and the query optimizer thinks it’ll be faster to just scan the entire clustered index.

In general this makes sense; SQL Server tries to pick plans that are good enough in most scenarios, and in general I think it chooses wisely.

However, sometimes SQL Server doesn’t pick great plans. Sometimes the plans it picks are downright terrible.

If that particular topic is interesting, I’ve a blog post from a few years back on a similar vein.

Grant Fritchey shows off the ShowPlan Schema:

Because the showplan schema contains notes throughout stating what the units of measure are, what each of the values means. For instance, I can explain why SerialDesiredMemory, DesiredMemory, RequestedMemory are identical:

…Provide memory grant estimate as well as actual runtime memory grant information. Serial required/desired memory attributes are estimated during query compile time for serial execution. The rest of attributes provide estimates and counters for query execution time considering actual degree of parallelism. SerialRequiredMemory: Required memory in KB if the query runs in serial mode. The query will not start without this memory. SerialDesiredMemory: Memory estimated to fit intermediate results in KB if the query runs in serial mode. RequiredMemory: Required memory in KB for the chosen degree of parallelism. If the query runs in serial mode, this is the same as SerialRequiredMemory. …

That’s taken directly from the 2017 schema. The units of measure are KB.

I’d never seen this before, so that’s going on my to-read list.

Bert Wagner takes a look at one of the lesser appreciated tricks in performance tuning:

I had a great question submitted to me (thank you Brandman!) that I thought would make for a good blog post:

…I’ve been wondering if it really matters from a performance standpoint where I start my queries. For example, if I join from A-B-C, would I be better off starting at table B and then going to A & C?

The short answer: Yes. And no.

One of my favorite query tuning books is SQL Tuning by Dan Tow.  Parts of it are rather dated at this point—like pretty much anything involving a rule-based optimizer—but the gist still works well.  What it comes down to is finding the best single table from which to drive your query (based on table size, filters, etc.) and selecting the appropriate join order afterward.  It’s fairly time-consuming effort, but for the 0.5-1% of queries which really need it, it can be the difference between an awful plan and a good one.

Arun Sirpal gives us some helpful information regarding interleaved execution in SQL Server 2017:

I have read-only T-SQL that references the MSTVF. I did have some code that use both data modifications and cross apply but interleaved execution does not occur in those scenarios.

So on my SQL Server 2017 instance I set the database to 110 compatibility mode and set query store on where then I execute my code.

Note that 110 is the compatibility mode for SQL Server 2012.  That becomes an important part of Arun’s story.

Lukas Eder explains one potential issue with window functions against large data sets:

Usually, this blog is 100% pro window functions and advocates using them at any occasion. But like any tool, window functions come at a price and we must carefully evaluate if that’s a price we’re willing to pay. That price can be a sort operation. And as we all know, sort operations are expensive. They follow O(n log n) complexity, which must be avoided at all costs for large data sets.

In a previous post, I’ve described how to calculate a running total with window functions (among other ways). In this post, we’re going to calculate the cumulative revenue at each payment in our Sakila database.

This is a good article comparing how different RDBMS products handle a fairly complicated windowed query and what you can do to improve performance.