Hi, I plan to create partitions for one large table based on date so I can load data quicker in SSIS. I am new to partition by the way. I have couple questions: 1. I understand that we can do "sliding window" to load data and get rid of or archive old data. Since we need to query all history data all the time, is it ok to keep everything in stead of sliding old data out (If I am using the right term). 2. Ideally we don't touch "old" partitions once they are loaded if I understand partition correctly, but what if data come in do contain some history data so they need to be insert into those history partitions. What do I need to do? or would this cause any problem? 3. If I load everything into a staging table and add to the big partitioned table later, what if the staging table contains data I mentioned in question #2? Would DB engine automatically organize data and insert them into different partitions based on date Thanks

This is right place to start http://weblogs.sqlteam.com/dang/archive/2008/09.aspx

Thanks, I will take a look.

This partition's introduction article by Kimberly Tripp is still the best reference.SSIS Tasks Components Scripts Services | http://www.cozyroc.com/

See an good walk trough on horizontal partitioning for archival here: http://www.mssqltips.com/tip.asp?tip=1796 The beauty of that approach is in "automated" insertion of the historical data into the proper partition. Arthur My Blog