Partition

SQLServerCentral is supported by Red Gate Software Ltd.

Popular Topics

Home

Members

Calendar

Who's On

Home » SQL Server 2005 » Development » Partition

Partition

Rate Topic

Display Mode

Topic Options

Author

Message

ningaraju.n

Posted Thursday, August 19, 2010 12:54 AM

SSC-Enthusiastic

Group: General Forum Members
Last Login: Thursday, February 21, 2013 3:14 AM
Points: 134, Visits: 425

Hi All,

I have a table which has around 50 million records and its increasing exponentially, and it stores the information of user and his attended question details for eg:sample data looks like this
ScheduleUserId,QuestionId,MaxScore,UserScore,IsCorrect,HintUsed
2345 32 2 2 1 0
2345 33 2 0 0 0

etc

we thought of doing the partition please provide some good approaches for this .

Post #971642

Stewart "Arturius" Campbell

Posted Tuesday, August 24, 2010 5:05 AM

Hall of Fame

Group: General Forum Members
Last Login: Yesterday @ 1:49 AM
Points: 3,123, Visits: 4,310

Question to consider:
> which column is likely to be used in range-based searches (usually the column with the clustered index)?
This would be the column to effect the partition on, e.g. using range left for customer for values 1000, 2000, 3000 etc
> how many filegroups and physical files would be required (to split the data accross numerous drives, if applicable)?

____________________________________________
Space, the final frontier? not any more...
All limits henceforth are self-imposed.
“libera tute vulgaris ex”

Post #974014

ningaraju.n

Posted Wednesday, August 25, 2010 12:21 AM

SSC-Enthusiastic

Group: General Forum Members
Last Login: Thursday, February 21, 2013 3:14 AM
Points: 134, Visits: 425

Thanks stewart,

We are using the scheduleuserid in most of the cases in where clause and sometimes questionId along with that.
and currently we have only one file group .
please suggest.

Post #974649

Stewart "Arturius" Campbell

Posted Wednesday, August 25, 2010 12:45 AM

Hall of Fame

Group: General Forum Members
Last Login: Yesterday @ 1:49 AM
Points: 3,123, Visits: 4,310

ningaraju.n (8/25/2010)

Thanks stewart,

We are using the scheduleuserid in most of the cases in where clause and sometimes questionId along with that.
and currently we have only one file group .
please suggest.

based on you reply, I would recommend
> creating at least one new filegroup (with associated physical file);
> build the partition function using the column most frequently used in range-based searches (scheduleuserid), using, for example, the ranges as shown in my first post on this subject;
> create the partition scheme to link the partition function to the new filegroup;
> build the table on the new partition scheme;
> create the clustered index on the partitioned column, also on the new partition scheme;
> Refer to books online with regard to the rules governing indexes on partitioned tables: http://msdn.microsoft.com/en-US/library/ms188706(v=SQL.90).aspx

or Kimberly Tripp's article: http://msdn.microsoft.com/en-us/library/ms345146(SQL.90).aspx

____________________________________________
Space, the final frontier? not any more...
All limits henceforth are self-imposed.
“libera tute vulgaris ex”

Post #974656

Marcos Leandro Rosa

Posted Wednesday, August 25, 2010 3:34 AM

Grasshopper

Group: General Forum Members
Last Login: Thursday, July 12, 2012 12:03 PM
Points: 17, Visits: 70

Hi,
Just a matter of complementing our friend's answer:

How many disk do you have available to spread your files?
Why do you want to split your table? [Is it for increasing performance?]::[Is it a matter of space?]

You have also to create a Partitioned Index and not just common index to get best performance.

First of all you have to measure why you want to have the database split then we can go with further investigation.

You can see my youtube channel I have a video talking about partitioned tables
http://www.youtube.com/watch?v=HDx8XCoAw3w

Best Regards,
Marcos Rosa / marcosfac@gmail.com

Post #974707

ningaraju.n

Posted Wednesday, August 25, 2010 4:06 AM

SSC-Enthusiastic

Group: General Forum Members
Last Login: Thursday, February 21, 2013 3:14 AM
Points: 134, Visits: 425

Thanks stewart,

I understand most of your points except the below point
>build the table on the new filegroup
I have to do partition for existing table

and can you please suggest the way to do that point('build the table on the new filegroup')

Post #974717

ningaraju.n

Posted Wednesday, August 25, 2010 4:24 AM

SSC-Enthusiastic

Group: General Forum Members
Last Login: Thursday, February 21, 2013 3:14 AM
Points: 134, Visits: 425

Thanks Rosa,

I don't know about the availability of disks,but we want partition to improve the performance

Please suggest in this regard

Post #974722

Stewart "Arturius" Campbell

Posted Wednesday, August 25, 2010 4:36 AM

Hall of Fame

Group: General Forum Members
Last Login: Yesterday @ 1:49 AM
Points: 3,123, Visits: 4,310

ningaraju.n (8/25/2010)

There are 2 methods of doing so.
Please refer to http://msdn.microsoft.com/en-US/library/ms175864(v=SQL.90).aspx for details (specifically the section regarding Converting a Nonpartitioned Table to a Partitioned Table)

____________________________________________
Space, the final frontier? not any more...
All limits henceforth are self-imposed.
“libera tute vulgaris ex”

Post #974725

antonyrajcletus

Posted Wednesday, December 08, 2010 3:43 AM

Forum Newbie

Group: General Forum Members
Last Login: Tuesday, April 12, 2011 8:17 AM
Points: 1, Visits: 7

Hi

Actually I have a requirement as below

1) To have a database in sql server 2008 with a set of tables (which will have the fields snapshotdate and frequency along with other data fields)
2) This set of tables should maintain only the latest 2 daily snapshot dates of data and latest 3 monthly snapshot dates of data and the older one should be removed everytime the new data comes in.

--> Please let me know the complete steps of how we can do it through partitions in SQL server 2008 which I require very urgently

Thanks,
Cletus

Post #1031760

« Prev Topic | Next Topic »

Permissions