SQL Server 2008 : Indexing for Performance - Putting It All Together (part 2) - Clustered Index Seeks

3. Clustered Index Seeks

The first two index creations in Listing 1 create clustered indexes on the SalesOrderHeader and Product tables. Then comes a query. Then the listing creates a clustered index on the SalesOrderDetail table. Finally, the query is executed again, showing the improvement from using a clustered index over the table scan.

Example 1. SQL Script to Create Clustered Indexes and Retrieve Data from the Newly Created Tables

CREATE CLUSTERED INDEX ix_SalesOrderId ON
apWriter.SalesOrderHeader(SalesOrderId)

CREATE CLUSTERED INDEX ix_ProductId ON apWriter.Product(ProductId)

SELECT *
FROM apWriter.SalesOrderDetail
WHERE SalesOrderId = 74853

CREATE CLUSTERED INDEX ix_SalesOrderIdDetailId
ON apWriter.SalesOrderDetail(SalesOrderId,SalesOrderDetailId)

SELECT *
FROM apWriter.SalesOrderDetail
WHERE SalesOrderId = 74853

Figure 1 shows the execution plans from executing Listing 1. Notice the missing index message. You are notified of missing indexes by default in SQL Server 2008. The information in that message is the same that you could get manually by querying the missing index views described earlier.

Figure 3. The execution plan of a table scan versus a clustered index seek

Also notice the cost difference between the table scan and the clustered index seek. If you run your multiple table query for a sales order, you will see that clustered index seeks are used instead of table scans. For example, execute the following query. You should see an execution plan like that shown in Figure 2.

SELECT soh.SalesOrderId, soh.OrderDate, soh.ShipDate,p.Name, sod.OrderQty,
 sod.UnitPrice, sod.LineTotal
FROM apWriter.SalesOrderHeader soh
 JOIN apWriter.SalesOrderDetail sod ON soh.SalesOrderId = sod.salesOrderId
 JOIN apWriter.Product p ON sod.ProductId = p.ProductId
WHERE soh.SalesOrderId = 74853

Figure 4. The execution plan of joining multiple tables using clustered index seeks