Partition Question: Archiving Data

Question

I'm purging and archiving data from a table based on date range. During purge process I'm merging partition function boundary points and then removing partition file.
I was wondering what will going to happen if user wants to enter the data which falls under boundary range that I already purged.
For example: I purged data older than 2010 and remove partition file "Inventory_201006" after merging partition function range. After that business user wants to enter data that is for 2008-06-01. Wondering where It would go.
PS: I tried entering older data but got an error:
"The filegroup "Inventory_200806" has no files assigned to it. Tables, indexes, text columns, ntext columns, and image columns cannot be populated on this filegroup until a file is added."
Do I need to mention default partition file group in partition scheme or somewhere else before removing file , so any data older than 2010 should be inserted in that default file group?
Thanks!
ZK

Answer 1

Partition always ranges from negative infinity to positive infinity. You apply boundary points through partition function which are applied to table through partition scheme which are applied on file groups.
so, even if you merge the boundary points,  the overall range will not change but the data will move between partitions. Also, instead of dropping the file and file group you could potentially use that as the next file group as you merge the partitions.
 try this example:
--create database with all filegroups needed
CREATE DATABASE [myPartition]
CONTAINMENT = NONE
ON PRIMARY
( NAME = N'myPartition', FILENAME = N'C:\Program Files\Microsoft SQL Server\MSSQL11.SQL2012\MSSQL\DATA\myPartition.mdf' , SIZE = 4096KB , FILEGROWTH = 1024KB ),
FILEGROUP [FG1]
( NAME = N'FG1', FILENAME = N'C:\Program Files\Microsoft SQL Server\MSSQL11.SQL2012\MSSQL\DATA\FG1.ndf' , SIZE = 4096KB , FILEGROWTH = 1024KB ),
FILEGROUP [FG2]
( NAME = N'FG2', FILENAME = N'C:\Program Files\Microsoft SQL Server\MSSQL11.SQL2012\MSSQL\DATA\FG2.ndf' , SIZE = 4096KB , FILEGROWTH = 1024KB ),
FILEGROUP [FG3]
( NAME = N'FG3', FILENAME = N'C:\Program Files\Microsoft SQL Server\MSSQL11.SQL2012\MSSQL\DATA\FG3.ndf' , SIZE = 4096KB , FILEGROWTH = 1024KB ),
FILEGROUP [FG4]
( NAME = N'FG4', FILENAME = N'C:\Program Files\Microsoft SQL Server\MSSQL11.SQL2012\MSSQL\DATA\FG4.ndf' , SIZE = 4096KB , FILEGROWTH = 1024KB )
LOG ON
( NAME = N'myPartition_log', FILENAME = N'C:\Program Files\Microsoft SQL Server\MSSQL11.SQL2012\MSSQL\DATA\myPartition_log.ldf' , SIZE = 4096KB , FILEGROWTH = 10%)
GO
USE [myPartition]
GO
IF NOT EXISTS (SELECT name FROM sys.filegroups WHERE is_default=1 AND name = N'PRIMARY')
ALTER DATABASE [myPartition] MODIFY FILEGROUP [PRIMARY] DEFAULT
GO
-- create partition function
Create Partition Function pf_Date(date)
as Range RIGHT For Values ('01/01/2010','01/01/2011','01/01/2012')
--create partition scheme
Create Partition Scheme ps_Date
as partition pf_Date
to (FG1,FG2,FG3,FG4)
/*--at this point, your boundary points are :
-infinity to 12/31/2009 - FG1
01/01/2010 to 12/31/2010 - FG2
01/01/2011 to 12/31/2011 - FG3
01/01/2012 to infinity - FG4 */
--create table with this partition
Create Table mytable(sno int identity(1,1),sname varchar(20),sdate date constraint PK_mytable Primary Key clustered (sdate)) on Ps_Date(sDate)
--insert some data
declare @a date
set @a='01/01/2009'
while (@a<'01/01/2014')
begin
insert into mytable(sname,sdate)
select cast(year(@a) as varchar(4))+'part',@a
set @a=dateadd(day,1,@a)
end
--check the row counts in each partition
select row_count,* from sys.dm_db_partition_stats where object_id=object_id('mytable')
select A.name [PartitionScheme],B.name as [PartitionFunction],C.boundary_id,C.value
from sys.partition_schemes A
left outer join sys.partition_functions B on A.function_id=b.function_id
left outer join sys.partition_range_values C on C.function_id=B.function_id
--merge the partition
ALTER PARTITION FUNCTION pf_Date ()
MERGE RANGE ('01/01/2010')
---- or otherwise,you could reuse the files/filegroup
ALTER PARTITION SCHEME ps_Date
NEXT USED FG2
ALTER PARTITION FUNCTION pf_Date ()
split RANGE ('01/01/2013')
/* at this point the boundary points are
-infinity to 12/31/2010 - FG1
01/01/2011 to 12/31/2011 - FG3
01/01/2012 to 12/31/2012 - FG4
01/01/2013 to infinity - FG2*/
--check the row counts in each partition
select row_count,* from sys.dm_db_partition_stats where object_id=object_id('mytable')
i assume, you understand that infinty is not really infinity but the minimum/maximum possible value for the data type.
Hope it Helps!!