The correct cardinality estimation using table variable

Hi Friends,

In my previous blog post, we have seen that SQL Sever estimates number of rows in a table variable was 1. Now the question is, Is there any way to make the correct estimation for the number of rows while using table variable?

My today’s blog post is focused on the correct cardinality estimation using table variable. The answer of the above question is: yes, the optimizer can make a correct cardinality estimation for table variables by using OPTION (RECOMPILE) hint for that. Using this hint also has some cons like new plan will generate every time. It’s like a trade-off between both the mechanism. Let me show you the same thing.

First: Use of table variable without Option Recompile

Declare @TV_SalesOrderDetail Table
(
	[SalesOrderID] [int] NOT NULL,
	[SalesOrderDetailID] [int] NOT NULL,
	[CarrierTrackingNumber] [nvarchar](25) NULL,
	[OrderQty] [smallint] NOT NULL,
	[ProductID] [int] NOT NULL,
	[SpecialOfferID] [int] NOT NULL,
	[UnitPrice] [money] NOT NULL,
	[UnitPriceDiscount] [money] NOT NULL,
	[LineTotal]  [money],
	[rowguid] [uniqueidentifier] ROWGUIDCOL,
	[ModifiedDate] [datetime] NOT NULL
)

INSERT INTO @TV_SalesOrderDetail
SELECT [SalesOrderID]
      ,[SalesOrderDetailID]
      ,[CarrierTrackingNumber]
      ,[OrderQty]
      ,[ProductID]
      ,[SpecialOfferID]
      ,[UnitPrice]
      ,[UnitPriceDiscount]
      ,[LineTotal]
      ,[rowguid]
      ,[ModifiedDate]
  FROM [Sales].[SalesOrderDetail]

SET STATISTICS TIME ON

Select SOD.UnitPrice,
SOH.DueDate, SOH.OrderDate
from @TV_SalesOrderDetail SOD
join [Sales].[SalesOrderHeader] SOH
on sod.SalesOrderID=SOH.SalesOrderID
Where SOH.OrderDate between '2011-07-01' AND '2011-07-31'
GO

Second: Use of table variable with Option Recompile

Declare @TV_SalesOrderDetail Table
(
	[SalesOrderID] [int] NOT NULL,
	[SalesOrderDetailID] [int] NOT NULL,
	[CarrierTrackingNumber] [nvarchar](25) NULL,
	[OrderQty] [smallint] NOT NULL,
	[ProductID] [int] NOT NULL,
	[SpecialOfferID] [int] NOT NULL,
	[UnitPrice] [money] NOT NULL,
	[UnitPriceDiscount] [money] NOT NULL,
	[LineTotal]  [money],
	[rowguid] [uniqueidentifier] ROWGUIDCOL,
	[ModifiedDate] [datetime] NOT NULL
)

INSERT INTO @TV_SalesOrderDetail
SELECT [SalesOrderID]
      ,[SalesOrderDetailID]
      ,[CarrierTrackingNumber]
      ,[OrderQty]
      ,[ProductID]
      ,[SpecialOfferID]
      ,[UnitPrice]
      ,[UnitPriceDiscount]
      ,[LineTotal]
      ,[rowguid]
      ,[ModifiedDate]
  FROM [Sales].[SalesOrderDetail]

SET STATISTICS TIME ON

Select SOD.UnitPrice,
SOH.DueDate, SOH.OrderDate
from @TV_SalesOrderDetail SOD
join [Sales].[SalesOrderHeader] SOH
on sod.SalesOrderID=SOH.SalesOrderID
Where SOH.OrderDate between '2011-07-01' AND '2011-07-31'
OPTION (RECOMPILE)
GO

From the above image you can see both actual and estimated number of rows are same and decision of using join also has been changed, the earlier nested loop was used but now it is using hash match.

Is there really a performance improvement due to the above change? Yes, you can see the stats below:

Stats for without Option Recompile

SQL Server Execution Times:

   CPU time = 281 ms,  elapsed time = 399 ms.

SQL Server parse and compile time:

   CPU time = 0 ms, elapsed time = 0 ms.

 SQL Server Execution Times:

   CPU time = 0 ms,  elapsed time = 0 ms.

Stats for with Option Recompile

SQL Server Execution Times:

   CPU time = 0 ms,  elapsed time = 156 ms.

SQL Server parse and compile time:

   CPU time = 0 ms, elapsed time = 0 ms.

 SQL Server Execution Times:

   CPU time = 0 ms,  elapsed time = 0 ms.

HAPPY LEARNING!

Regards:

Prince Kumar Rastogi

Like us on FaceBook | Join the fastest growing SQL Server group on FaceBook

Follow Prince Rastogi on Twitter | Follow Prince Rastogi on FaceBook