Saturday, 24 May 2014

Is there any differences between Table Scan and Index Scan

Introduction


As we all know that the developer prefers the Index Seek in the case of performance of a query and we all know that what is Index Seek and who it improve the performance. In this article we are not going to discuss about Index Seek.

A table scan is performed on a table which does not have an Index upon it (a heap) – it looks at the rows in the table and an Index Scan is performed on an indexed table – the index itself.

Here we are trying discuss about a specified scenario occurs related to Table Scan and Index Scan.

Scenario Description

Suppose we have a table object without any Index on it. The name of the Table Object is tbl_WithoutIndex and another Table Object called tbl_WuthIndex

-- Heap Table Defination
IF OBJECT_ID(N'dbo.tbl_WithoutIndex', N'U') IS NOT NULL
   BEGIN
      DROP TABLE [dbo].[tbl_WithoutIndex];
   END
GO

CREATE TABLE [dbo].[tbl_WithoutIndex]
       (
         EMPID    INT         NOT NULL,
         EMPNAME  VARCHAR(50) NOT NULL
       );
GO

-- Insert Some Records
INSERT INTO  [dbo].[tbl_WithoutIndex]
       (EMPID, EMPNAME)
VALUES (101, 'Joydeep Das'),
       (102, 'Sukamal Jana'),
       (103, 'Ranajit Shinah');
GO      

-- Table with Index (Clustered Index for primary Key)                        
IF OBJECT_ID(N'dbo.tbl_WithIndex', N'U') IS NOT NULL
   BEGIN
      DROP TABLE [dbo].[tbl_WithIndex];
   END
GO

CREATE TABLE [dbo].[tbl_WithIndex]
       (
         EMPID    INT         NOT NULL PRIMARY KEY,
         EMPNAME  VARCHAR(50) NOT NULL
       );
GO

-- Finding Index Name
sp_helpindex tbl_WithIndex;

index_name
index_description
index_keys
PK__tbl_With__14CCD97D3AD6B8E2
clustered, unique, primary key located on PRIMARY
EMPID

-- Insert Some Records
INSERT INTO  [dbo].[tbl_WithIndex]
       (EMPID, EMPNAME)
VALUES (101, 'Joydeep Das'),
       (102, 'Sukamal Jana'),
       (103, 'Ranajit Shinah');
GO

Now we are going to compare the execution plan output

SELECT * FROM [dbo].[tbl_WithoutIndex];      



As we can see in the Execution plan, there is Table Scan. As the table has no Index (Here I mean the clustered Index) the table is a Heap. So when we put the SELECT statement the entire table scanned.

SELECT * FROM [dbo].[tbl_WithIndex];



Here he Table has a PRIMARY KEY, so it has a Clustered Index on it. But here we are not putting the Index columns on the WHERE clause, so the Clustered Index Scan Occurs.

Close Observation of Execution Plan



Please remember that the table has small number of records.

Table Name
Estimated IO Cost
tbl_WithoutIndex
0.0032035
Tbl_WithIndex
0.003125

If we see the Estimated Operation Cost, it would be same for both the Query (0.0032853).

Question in Mind

Here we can see the Estimated Operation cost for both the Query is same. So Question is in the mind that, if a Index Scan occur we can drop the index and use the heap (in our case). So is there any other difference between them.

How they are Differences

Here we understand what the internal difference between Table Scan and Index Scan.
When the table scan occurs MS SQL server reads all the Rows and Columns into memory. When the Index Scan occurs, it's going to read all the Rows and only the Columns in the index.
Effects in Performance
In case of performance the Table Scan and Index Scan both have the same output, if we have use the single table SELECT statement. But it differs in the case of Index Scan when we are going to JOIN table.


Hope you like it.

Posted by: MR. JOYDEEP DAS

No comments:

Post a Comment