Sql-server – Multi-statement TVF vs Inline TVF Performance

set-returning-functionssql serversql-server-2012

Comparing some of the answers on the Palindrome question(10k+ users only, since I've deleted the answer), I'm getting confusing results.

I proposed a multi-statement, schema-bound TVF which I thought would be faster than running a standard function, which it is. I was also under the impression that the multi-statement TVF would be "inlined", although I am wrong on that count, as you'll see below. This question is about the performance difference of those two styles of TVF. First, you'll need to see the code.

Here is the multi-statement TVF:

IF OBJECT_ID('dbo.IsPalindrome') IS NOT NULL
DROP FUNCTION dbo.IsPalindrome;
GO
CREATE FUNCTION dbo.IsPalindrome
(
    @Word NVARCHAR(500)
) 
RETURNS @t TABLE
(
    IsPalindrome BIT NOT NULL
)
WITH SCHEMABINDING
AS
BEGIN
    DECLARE @IsPalindrome BIT;
    DECLARE @LeftChunk NVARCHAR(250);
    DECLARE @RightChunk NVARCHAR(250);
    DECLARE @StrLen INT;
    DECLARE @Pos INT;
    SET @RightChunk = '';
    SET @IsPalindrome = 0;
    SET @StrLen = LEN(@Word) / 2;
    IF @StrLen % 2 = 1 SET @StrLen = @StrLen - 1;
    SET @Pos = LEN(@Word);
    SET @LeftChunk = LEFT(@Word, @StrLen);
    WHILE @Pos > (LEN(@Word) - @StrLen)
    BEGIN
        SET @RightChunk = @RightChunk + SUBSTRING(@Word, @Pos, 1)
        SET @Pos = @Pos - 1;
    END
    IF @LeftChunk = @RightChunk SET @IsPalindrome = 1;
    INSERT INTO @t VALUES (@IsPalindrome);
    RETURN
END
GO

The inline-TVF:

IF OBJECT_ID('dbo.InlineIsPalindrome') IS NOT NULL
DROP FUNCTION dbo.InlineIsPalindrome;
GO
CREATE FUNCTION dbo.InlineIsPalindrome
(
    @Word NVARCHAR(500)
)
RETURNS TABLE
WITH SCHEMABINDING
AS RETURN (
    WITH Nums AS
    (
      SELECT
        N = number
      FROM
        dbo.Numbers
    )
    SELECT
      IsPalindrome =
        CASE
          WHEN EXISTS
          (
            SELECT N
            FROM Nums
            WHERE N <= L / 2
              AND SUBSTRING(S, N, 1) <> SUBSTRING(S, 1 + L - N, 1)
          )
          THEN 0
          ELSE 1
        END
    FROM
      (SELECT LTRIM(RTRIM(@Word)), LEN(@Word)) AS v (S, L)
);
GO

The Numbers table in the above function is defined as:

CREATE TABLE dbo.Numbers
(
    Number INT NOT NULL 
);

Note: The numbers table does not have any indexes and no primary key, and contains 1,000,000 rows.

A test-bed temporary-table:

IF OBJECT_ID('tempdb.dbo.#Words') IS NOT NULL
DROP TABLE #Words;
GO
CREATE TABLE #Words 
(
    Word VARCHAR(500) NOT NULL
);

INSERT INTO #Words(Word) 
SELECT o.name + REVERSE(w.name)
FROM sys.objects o
CROSS APPLY (
    SELECT o.name
    FROM sys.objects o
) w;

On my test system the above INSERT results in 16,900 rows being inserted into the #Words table.

To test the two variations, I SET STATISTICS IO, TIME ON; and use the following:

SELECT w.Word
    , p.IsPalindrome
FROM #Words w
    CROSS APPLY dbo.IsPalindrome(w.Word) p
ORDER BY w.Word;


SELECT w.Word
    , p.IsPalindrome
FROM #Words w
    CROSS APPLY dbo.InlineIsPalindrome(w.Word) p
ORDER BY w.Word;

I expected the InlineIsPalindrome version to be significantly faster, however the following results do not support that supposition.

Multi-statement TVF:

Table '#A1CE04C3'. Scan count 16896, logical reads 16900, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
Table 'Worktable'. Scan count 0, logical reads 0, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
Table '#Words'. Scan count 1, logical reads 88, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.

SQL Server Execution Times:
CPU time = 1700 ms, elapsed time = 2022 ms.
SQL Server parse and compile time:
CPU time = 0 ms, elapsed time = 0 ms.

Inline TVF:

Table 'Numbers'. Scan count 1, logical reads 1272030, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
Table 'Worktable'. Scan count 0, logical reads 0, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
Table '#Words'. Scan count 1, logical reads 88, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.

SQL Server Execution Times:
CPU time = 137874 ms, elapsed time = 139415 ms.
SQL Server parse and compile time:
CPU time = 0 ms, elapsed time = 0 ms.

The execution plans look like:

Why is the inline variant so much slower than the multi-statement variant, in this case?

In response to a comment by @AaronBertrand, I've modified the dbo.InlineIsPalindrome function to limit the rows returned by the CTE to match the length of the input word:

CREATE FUNCTION dbo.InlineIsPalindrome
(
    @Word NVARCHAR(500)
)
RETURNS TABLE
WITH SCHEMABINDING
AS RETURN (
    WITH Nums AS
    (
      SELECT
        N = number
      FROM
        dbo.Numbers
      WHERE 
        number <= LEN(@Word)
    )
    SELECT
      IsPalindrome =
        CASE
          WHEN EXISTS
          (
            SELECT N
            FROM Nums
            WHERE N <= L / 2
              AND SUBSTRING(S, N, 1) <> SUBSTRING(S, 1 + L - N, 1)
          )
          THEN 0
          ELSE 1
        END
    FROM
      (SELECT LTRIM(RTRIM(@Word)), LEN(@Word)) AS v (S, L)
);

As @MartinSmith suggested, I've added a primary key and clustered index to the dbo.Numbers table, which certainly helps and would be closer to what one would expect to see in a production environment.

Re-running the tests above now results in the following stats:

CROSS APPLY dbo.IsPalindrome(w.Word) p:

(17424 row(s) affected)
Table '#B1104853'. Scan count 17420, logical reads 17424, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
Table 'Worktable'. Scan count 0, logical reads 0, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
Table '#Words'. Scan count 1, logical reads 90, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.

SQL Server Execution Times:
CPU time = 1763 ms, elapsed time = 2192 ms.

dbo.FunctionIsPalindrome(w.Word):

(17424 row(s) affected)
Table 'Worktable'. Scan count 0, logical reads 0, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
Table '#Words'. Scan count 1, logical reads 90, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.

SQL Server Execution Times:
CPU time = 328 ms, elapsed time = 424 ms.

CROSS APPLY dbo.InlineIsPalindrome(w.Word) p:

(17424 row(s) affected)
Table 'Numbers'. Scan count 1, logical reads 237100, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
Table 'Worktable'. Scan count 0, logical reads 0, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
Table '#Words'. Scan count 1, logical reads 90, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.

SQL Server Execution Times:
CPU time = 17737 ms, elapsed time = 17946 ms.

I'm testing this on SQL Server 2012 SP3, v11.0.6020, Developer Edition.

Here is the definition of my numbers table, with the primary key and clustered index:

CREATE TABLE dbo.Numbers
(
    Number INT NOT NULL 
        CONSTRAINT PK_Numbers
        PRIMARY KEY CLUSTERED
);

;WITH n AS
(
    SELECT v.n 
    FROM (
        VALUES (1) 
            ,(2) 
            ,(3) 
            ,(4) 
            ,(5) 
            ,(6) 
            ,(7) 
            ,(8) 
            ,(9) 
            ,(10)
        ) v(n)
)
INSERT INTO dbo.Numbers(Number)
SELECT ROW_NUMBER() OVER (ORDER BY n1.n)
FROM n n1
    , n n2
    , n n3
    , n n4
    , n n5
    , n n6;

Best Answer

Your numbers table is a heap and is potentially being fully scanned each time.

Add a clustered primary key on Number and try the following with a forceseek hint to get the desired seek.

As far as I can tell this hint is needed as SQL Server just estimates that 27% of the table will match the predicate (30% for the <= and reduced down to 27% by the <>). And therefore that it will only have to read 3-4 rows before finding one that matches and it can exit the semi join. So the scan option is costed very cheaply. But in fact if any palindromes do exist then it will have to read the whole table so this is not a good plan.

CREATE FUNCTION dbo.InlineIsPalindrome
(
    @Word NVARCHAR(500)
)
RETURNS TABLE
WITH SCHEMABINDING
AS RETURN (
    WITH Nums AS
    (
      SELECT
        N = number
      FROM
        dbo.Numbers WITH(FORCESEEK)
    )
    SELECT
      IsPalindrome =
        CASE
          WHEN EXISTS
          (
            SELECT N
            FROM Nums
            WHERE N <= L / 2
              AND SUBSTRING(S, N, 1) <> SUBSTRING(S, 1 + L - N, 1)
          )
          THEN 0
          ELSE 1
        END
    FROM
      (SELECT LTRIM(RTRIM(@Word)), LEN(@Word)) AS v (S, L)
);
GO

With those changes in place it flies for me (takes 228ms)

Subqueries in CASE expressions

Consider the following (perfectly legal) query:

DECLARE @Base AS TABLE (a integer NULL);
DECLARE @When AS TABLE (b integer NULL);
DECLARE @Then AS TABLE (c integer NULL);
DECLARE @Else AS TABLE (d integer NULL);

SELECT
    CASE
        WHEN (SELECT W.b FROM @When AS W) = 1
            THEN (SELECT T.c FROM @Then AS T)
        ELSE (SELECT E.d FROM @Else AS E)
    END
FROM @Base AS B;

The semantics of CASE are that WHEN/ELSE clauses are generally evaluated in textual order. In the query above, it would be incorrect for SQL Server to return an error if the ELSE subquery returned more than one row, if the WHEN clause was satisfied. To respect these semantics, the optimizer produces a plan that uses pass-through predicates:

Pass-through predicates

The inner side of the nested loop joins are only evaluated when the pass-through predicate returns false. The overall effect is that CASE expressions are tested in order, and subqueries are only evaluated if no previous expression was satisfied.

CASE expressions with an EXISTS subquery

Where a CASE subquery uses EXISTS, the logical existence test is implemented as a semi-join, but rows that would normally be rejected by the semi-join have to be retained in case a later clause needs them. Rows flowing through this special kind of semi-join acquire a flag to indicate if the semi-join found a match or not. This flag is known as the probe column.

The details of the implementation is that the logical subquery is replaced by a correlated join ('apply') with a probe column. The work is performed by a simplification rule in the query optimizer called RemoveSubqInPrj (remove subquery in projection). We can see the details using trace flag 8606:

SELECT
    T1.ID,
    CASE
        WHEN EXISTS 
        (
            SELECT 1
            FROM #T2 AS T2
            WHERE T2.ID = T1.ID
        ) THEN 1 
    ELSE 0
    END AS DoesExist
FROM #T1 AS T1
WHERE T1.ID BETWEEN 5000 AND 7000
OPTION (QUERYTRACEON 3604, QUERYTRACEON 8606);

Part of the input tree showing the EXISTS test is shown below:

ScaOp_Exists 
    LogOp_Project
        LogOp_Select
            LogOp_Get TBL: #T2
            ScaOp_Comp x_cmpEq
                ScaOp_Identifier [T2].ID
                ScaOp_Identifier [T1].ID

This is transformed by RemoveSubqInPrj to a structure headed by:

LogOp_Apply (x_jtLeftSemi probe PROBE:COL: Expr1008)

This is the left semi-join apply with probe described previously. This initial transformation is the only one available in SQL Server query optimizers to date, and compilation will simply fail if this transformation is disabled.

One of the possible execution plan shapes for this query is a direct implementation of that logical structure:

NLJ Semi Join with Probe

The final Compute Scalar evaluates the result of the CASE expression using the probe column value:

Compute Scalar expression

The basic shape of the plan tree is preserved when the optimize considers other physical join types for the semi join. Only merge join supports a probe column, so a hash semi join, though logically possible, is not considered:

Merge with probe column

Notice the merge outputs an expression labelled Expr1008 (that the name is the same as before is a coincidence) though no definition for it appears on any operator in the plan. This is just the probe column again. As before, the final Compute Scalar uses this probe value to evaluate the CASE.

The problem is that the optimizer doesn't fully explore alternatives that only become worthwhile with merge (or hash) semi join. In the nested loops plan, there is no advantage to checking if rows in T2 match the range on every iteration. With a merge or hash plan, this could be a useful optimization.

If we add a matching BETWEEN predicate to T2 in the query, all that happens is that this check is performed for each row as a residual on the merge semi join (tough to spot in the execution plan, but it is there):

SELECT
    T1.ID,
    CASE
        WHEN EXISTS 
        (
            SELECT 1
            FROM #T2 AS T2
            WHERE T2.ID = T1.ID
            AND T2.ID BETWEEN 5000 AND 7000 -- New
        ) THEN 1 
    ELSE 0
    END AS DoesExist
FROM #T1 AS T1
WHERE T1.ID BETWEEN 5000 AND 7000;

Residual predicate

We would hope that the BETWEEN predicate would instead be pushed down to T2 resulting in a seek. Normally, the optimizer would consider doing this (even without the extra predicate in the query). It recognizes implied predicates (BETWEEN on T1 and the join predicate between T1 and T2 together imply the BETWEEN on T2) without them being present in the original query text. Unfortunately, the apply-probe pattern means this is not explored.

There are ways to write the query to produce seeks on both inputs to a merge semi join. One way involves writing the query in quite an unnatural way (defeating the reason I generally prefer EXISTS):

WITH T2 AS
(
    SELECT TOP (9223372036854775807) * 
    FROM #T2 AS T2 
    WHERE ID BETWEEN 5000 AND 7000
)
SELECT 
    T1.ID, 
    DoesExist = 
        CASE 
            WHEN EXISTS 
            (
                SELECT * FROM T2 
                WHERE T2.ID = T1.ID
            ) THEN 1 ELSE 0 END
FROM #T1 AS T1
WHERE T1.ID BETWEEN 5000 AND 7000;

TOP trick plan

I wouldn't be happy writing that query in a production environment, it's just to demonstrate that the desired plan shape is possible. If the real query you need to write uses CASE in this particular way, and performance suffers by there not being a seek on the probe side of a merge semi-join, you might consider writing the query using different syntax that produces the right results and a more efficient execution plan.

Sql-server – Which of these queries is best for performance

Sometimes I wonder if SHORT scripts really is the best thing to focus on.

The size of a script has little to do with how efficiently the query will execute. A more compact statement will likely consume fewer resources in terms of compilation, but (re)compilation is usually a rare occurrence in a live system.

Fewer table accesses is usually desirable, though, and this does lead to more compact code.

Very generally speaking, a smaller execution plan will yield better results, and a lower estimated cost will yield better results. Again, though, it's highly situational. Cost estimates in particular can be way off in some cases. It's important to measure the actual execution time, because at the end of the day, that's what matters.

With left joins i can achieve what i want with just a few lines. But then I tried with a longer script, using unions. Which is the best method?

First of all, we need to know how much data will be in these tables in a real system. Right now there's so little it will be difficult to use the STATISTICS TIME performance metrics to figure out a winner -- the results that come back will be dominated by factors other than the query execution. With more data, it's likely the plans will change, thus rendering the comparison here moot.

Having said that, by looking at the query plans as they are now from a logical point of view, the first one is the winner.

You can see that the Clustered Index Scan of quantities appears once in the first plan, while it appears four times in the second one. The second plan also contains an expensive Distinct Sort as a result of using UNIONs (this operator could be eliminated by using UNION ALLs instead, which won't change the results).

The first query could also probably be improved, by getting index seeks on the colors and sizes tables, instead of table scans. It might be worth trying a hash match plan as well (which is what you'll probably see when quantities and products are larger), but for tables this small, the startup cost may be too much overhead to be of benefit.

What I would suggest you do is run each of the statements you want to test 10,000+ times in a loop, figure out the average execution time, and then compare.

Best Answer

Related Solutions

Sql-server – Check existence with EXISTS outperform COUNT! … Not

Subqueries in CASE expressions

CASE expressions with an EXISTS subquery

Sql-server – Which of these queries is best for performance

Related Question