June 24, 2010 at 9:45 am
Scenario: Datawarehouse
SQL Server 2008 Enterprise sitting on Windows 2008 Server Enterprise.
I'm about to load a massive OLTP table of 400 M records, 500Gb in size, from Production Oracle 10g to a SQL Server DataWarehouse.
I have 2 SAN discs to put this table on, 500 GB each.
I'm planning to create two data files in one file group, one file on each SAN disk to achieve simple striping and improve Read performance.
My question is about indexes, that will be eventually built on this big table. I have one more SAN disk, 800 Gb in size,
which I'd like to use specifically for storing indexes. I'm aware that it's possible to place non-clustered indexes in a separate
file from data on a stand-alone disk, but I don't know how to implement this, where to start. Could anyone kindly point me to the right
documentation, or possible explain on a high level how to do it?
Thank you for your time,
Respectfully,
Sergei Z
June 24, 2010 at 1:48 pm
How about this approach:
1. create a second file group for the database, call it *Idx_group*
2. create one file there Idx_file, the file is sitting on a dedicated SAN *Index_SAN*
3. all non-clustered indexes will be created in this file group *Idx_group* by using this clause
CREATE INDEX...
...
[ ON { partition_scheme_name (column_name)
| filegroup_name
| default
}
]
<end>
Thanks for reading.
Sergei Z
Viewing 2 posts - 1 through 1 (of 1 total)
You must be logged in to reply to this topic. Login to reply