Sunday, October 5, 2008

Difference between ROW_NUMBER, RANK and DENSE_RANK

What is the Difference between ROW_NUMBER, RANK and DENSE_RANK? Which one to use?
This is very common question in the minds of SQL newbie's.
Lets take 1 simple example to understand the difference between 3.

First lets create some sample data :

-- create table
CREATE TABLE Salaries
(
Names VARCHAR(1),
SalarY INT
)
GO
-- insert data
INSERT INTO Salaries SELECT
'A',5000 UNION ALL SELECT
'B',5000 UNION ALL SELECT
'C',3000 UNION ALL SELECT
'D',4000 UNION ALL SELECT
'E',6000 UNION ALL SELECT
'F',10000
GO
-- Test the data
SELECT Names, Salary
FROM Salaries


Now lets query the table to get the salaries of all employees with their salary in descending order.
For that I'll write a query like this :

SELECT names
        , salary
        ,row_number () OVER (ORDER BY salary DESC) as ROW_NUMBER
        ,rank () OVER (ORDER BY salary DESC) as RANK
        ,dense_rank () OVER (ORDER BY salary DESC) as DENSE_RANK
FROM salaries


>>Output
NAMES SALARY ROW_NUMBER RANK DENSE_RANK
F 10000 1 1 1
E 6000 2 2 2
A 5000 3 3 3
B 5000 4 3 3
D 4000 5 5 4
C 3000 6 6 5

Interesting Names in the result are employee A, B and D. 
Row_number assign different number to them.
Rank and Dense_rank both assign same rank to A and B.
But interesting thing is what RANK and DENSE_RANK assign to next row?
Rank assign 5 to the next row, while dense_rank assign 4.

The numbers returned by the DENSE_RANK function do not have gaps and always have consecutive ranks.  The RANK function does not always return consecutive integers.  The ORDER BY clause determines the sequence in which the rows are assigned their unique ROW_NUMBER within a specified partition.

So question is which one to use?
Its all depends on your requirement and business rule you are following.
1. Row_number to be used only when you just want to have serial number on result set. It is not as intelligent as RANK and DENSE_RANK.
2. Choice between RANK and DENSE_RANK depends on business rule you are following. Rank leaves the gaps between number when it sees common values in 2 or more rows. DENSE_RANK don't leave any gaps between ranks.
So while assigning the next rank to the row RANK will consider the total count of rows before that row and DESNE_RANK will just give next rank according to the value.
So If you are selecting employee’s rank according to their salaries you should be using DENSE_RANK and if you are ranking students according to there marks you should be using RANK(Though it is not mandatory, depends on your requirement.)

Mangal Pardeshi

SQL MVP

Thursday, October 2, 2008

Oracle Database Software

Download from following links
Oracle Database 11g Release 1 (11.1.0.6.0)

Oracle Database 10g Release 1

Oracle9i Database

Mangal Pardeshi

SQL MVP

Microsoft SQL Server Trial Download Links

Download the SQL Server 2008/2005 180-day Trial Software.
Click on the respective links.

Microsoft® SQL Server® 2008 Enterprise Evaluation: Trial

Microsoft SQL Server 2005 Enterprise Evaluation Edition: Trial



Mangal Pardeshi

SQL MVP