Showplan Operator of the Week – Row Count Spool

Comments 0

Share to social media

1076-FA1.jpg

Over the past few weeks, we’ve featured the Spool operations (Table Spool(Eager Spool), Table Spool(Lazy Spool)and NonClustered Index Spool), and this week we’ll be completing the set with the Row Count Spool Showplan operator.

Of all spool operators we’ve seen, I think this is the most simple. The Row Count Spool operator just scans an input, counting how many rows are present, and returns the number of rows without any of the data they contained. This operator is used when it is important to check for the existence of rows, but not what data they hold. For example, if a Nested Loops operator performs a left anti semi join operation, and the join predicate applies to the inner input, a row count spool may be placed at the top of that input to cache the number of rows which satisfy the argument. Then, the Nested Loops operator can just use that row count information (because the actual data from the inner input is not needed) to determine whether to return the outer row or not (or rather, how many rows to return).

I know that’s a little tricky to wrap your head around at first without a concrete example, so to illustrate this behavior I’ll start, as always, by creating a table called “Pedidos” (which means Orders in Portuguese). The following script will create a table and populate it with some garbage data:

This is what the data looks like.

1076-FA2.jpg

Figure 1. The demonstration table and data.

Next, we’ve got a sample query which will return the ID and Value for all orders only if there is no order which was placed on 2009-01-01 for more than 10 items (i.e. Quantity (Quantidade) > 10).

For the query above, we have the following execution plan:

1076-FA3.jpg

Figure 2. The graphical execution plan, clearly employing the Row Count Spool

As we can see, this plan is using the operator Row count Spool to check the EXISTS Sub Query, which returns just a true (some row’s exist) or false (no rows exist) value, and so the actual contents of the rows doesn’t matter (do you see where I’m going with this?) Let’s take a closer look at what is happening.

First, the Clustered Index Scan reads all the rows in Pedido table using the ix index, after which all rows are passed to the Nested Loops Operator to be joined with the SubQuery result. Bear in mind that this Loops operator is working as a Left anti Semi Join, which means that only one side (the outer side) of the join will be returned. In our plan, this means that only the rows read from Ped1 will be returned to the Select operator, but for each row read in Ped1, the loop will look at the Row Count Spool to see if the subquery value exists (i.e. does the spool return a row or not).

SQL Server uses the Row Count Spool operator to avoid reading the Pedido (Ped2) table over and over again. It calls the Clustered Index Scan at the bottom of the execution plan, which returns just one row (or not, if none satisfy the subquery). This row (or rather, it’s existence) is then cached in the Spool, and this cached value is reused for each row of the Ped1 table in the nested loop.

Why is this spool even necessary? Let’s look at a comparison between a plan which uses the row count spool and a plan which doesn’t. As I did in my last article, I’ve written an XML plan to force the Query Optimizer to use my (Row-Count-spool-less) plan, and you can download the query here.

1076-FA4.jpg

Figure 3. The two comparison queries we’re about to run, one using the row count spool, and one not.

For the queries above, we have the following execution plans:

1076-FA5.jpg

Figure 4. The comparison execution plans, illustrating the huge difference made by the row count spool operator.

Take a look at the differences between the costs of two queries; that seems like pretty strong evidence for the usefulness of the Row Count Spool operator, doesn’t it? By avoiding having to read the Pedido table for each row of the Inner input to the nested loop, SQL Server creates a huge performance gain. Just look at the IO results for the two queries above:

I’ve nothing else to say about that, because I don’t think there’s much I can add to that kind of compelling evidence. That’s all folks, I hope you’ve enjoyed learning about Spools, and I’ll see you next week with more “Showplan Operators”.

If you missed last week’s thrilling Showplan Operator, you can see it here

Load comments

About the author

Fabiano Amorim

See Profile

Fabiano Amorim is a Data Platform MVP since 2011 that loves to conquer complex, challenging problems—especially ones that others aren’t able to solve. He first became interested in technology when his older brother would bring him to his work meetings at the age of 14. With over a decade of experience, Fabiano is well known in the database community for his performance tuning abilities. When he isn’t working, he loves to read and spend time with his family.

Fabiano Amorim's contributions