Updating SQL partition schemes & partition functions

My client's IT dept had set up SQL Server 2005 filegroups on a particular database for each fiscal quarter: fg2006Q4, fg2007Q1, fg2007Q2, ... up to fg2010Q4. I noticed today that the fg2010Qx files had not been used, still at 1MB each.

Upon further digging, I found the Partition Schemes and Partition Functions folders in Mgmt Studio.

One Scheme: CREATE PARTITION SCHEME [psEvent] AS PARTITION [pfEvent] TO ([fg2006Q4], [fg2007Q1], [fg2007Q2], [fg2007Q3], [fg2007Q4], [fg2008Q1], [fg2008Q2], [fg2008Q3], [fg2008Q4], [fg2009Q1], [fg2009Q2], [fg2009Q3], [fg2009Q4])

Corresponding Function: CREATE PARTITION FUNCTION [pfEvent](int) AS RANGE RIGHT FOR VALUES (1000300, 5493523, 25712235, 45243827, 67741212, 88804361, 106935897, 121343754, 122508529, 122512053, 139523161, 154878083)

Q1: I'd like to create filegroups for fg2011Qx... then can I simply ALTER PARTITION SCHEME with fg2010Q1 through fg2011Q4 added to the list?

Q2: I believe I understand the function ranges, but wondering where/how these ID's would have been generated (likely)? And more fundamentally, how would SQL know when to move onto the next filegroup (assuming it had existed when needed)?

Q3: Is there a safe & easy way to "re-balance" the filegroups... so that fg2010Qx's are used, and the fg2009Q4's aren't so big. I suppose not a huge deal if not, but it'd look cleaner if so.

Thanks,
Jeff
jdallen75Asked:
Who is Participating?
 
Sander StadSysteemontwikkelaar, Database AdministratorCommented:
Q2:
The numbers in the function are not ID's but are the boundaries of rows. This website shows you what you want to know: http://msdn.microsoft.com/en-us/library/ms187802(v=SQL.90).aspx

Q3: What do you mean by rebalance the filegroups? Do you want to change the number of records that are saved in the different filegroups?
0
 
Sander StadSysteemontwikkelaar, Database AdministratorCommented:
You can add an additional filegroup to the scheme with the following command:

ALTER PARTITION SCHEME psEvent
NEXT USED fg2011Qx;
GO

Than you can use the following script to alter the partition function:

IF EXISTS (SELECT * FROM sys.partition_functions
    WHERE name = 'pfEevent')
DROP PARTITION FUNCTION pfEvent;
GO
CREATE PARTITION FUNCTION pfEevent (int)
AS RANGE RIGHT FOR VALUES  (1000300, 5493523, 25712235, 45243827, 67741212, 88804361, 106935897, 121343754, 122508529, 122512053, 139523161, 154878083, xxxxxxx);
GO
0
 
jdallen75Author Commented:
Sstad,

Q3: I mean that, for instance, the files in fg2009Q1, Q2, and Q3 each total 10GB. However, since fg2009Q4 was the last listed in the partition scheme, it has been used since that point in time, and is now 80GB. fg2010Qx files are basically empty because they have not been used this year.

Thanks on the other points - I will look into them.

JA
0
Get your problem seen by more experts

Be seen. Boost your question’s priority for more expert views and faster solutions

 
jdallen75Author Commented:
I think I understand now... it looks as though the boundary values in the partition function list are SSID's of the rows in the tables. That part isn't all that intuitive, and more importantly does not cleanly divided up the data by calendar quarters. What seems to be cleaner would be to choose a datetime column from the table, and define its boundaries accordingly. "20090101, 20090401, 20090701..." is a heck of a lot clearer!
0
 
Sander StadSysteemontwikkelaar, Database AdministratorCommented:
That could work but make sure your column is in date or int format because I don't know how it works with text in this case.
0
 
jdallen75Author Commented:
I discovered the common problem in that the partition column has to be part of the primary key, which in this case it isn't. There are a couple of workaround solutions, be we've decided in this case (since it is an archive DB), to go with one partition over several filegroups.

Thanks for your guidance.
JA
0
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.