I’m trying to back up a bunch of computational data – about 16 – 18 TB at the moment.  I’ll be backing a single fileserver up to a ~36 TB RAID array connected to the server via fiber channel.

We add new data constantly but most of the data is pretty much static once it has been on the system for a week or so. Computational results get dropped on the server as they are created but these files sometimes will be deleted and the data re-run as the models are developed. Once the data has been there for a while it is less likely to change.

I’m considering how often I do a Full backup since even with fiber channel the backup is going to take a long time. The typical Full on the first weekend of the month, Differential on the other weekends and Incrementals every night seems like a lot.

One plan I’m looking at is a Full backup quarterly – starting at 21:01 am on the first Saturday after Dec 1^st, March 1^st, June 1^st, and September 1^st.

Differentials would then be done every intervening month  and incremental every day. I would love to archive data older than 1 year off onto another medium but have no facilities for that at the moment.

First, are there any red flags I don’t see in this scheme and second, how do I write the schedule to do this?

Thanks,

Chris

J. Chris Johnston
Polymers Branch, Structures and Materials Division   (216) 433-5029
NASA John H. Glenn Research Center                              (216) 977-7132 (fax)
21000 Brookpark Rd MS 49-3                              James.C.Johnston-1 AT nasa DOT gov
Cleveland, Ohio 44135
                    

NOTICE: Lack of planning on your part does not constitute an emergency on my part.