NDMP differential incremental backup

Just  to make sure I give all of you a clearer idea of what I intend to do, here is an example of what I do now and what I would like to achieve :

My goal : I would like to make use of the mostly unused CDL B (in a separate DR site) and keep my daily incremental images on this CDL instead of having to physically vault them to physical tapes for a month's retention. I would like to cut down on the use of physical tapes. Also, I would like to relieve some burden on the NAS A (production) by doing daily incremental backups from NAS B (NAS A's snapmirror destination/DR). NAS B and CDL B are located on the same site/DC. NAS A and CDL A are in another site/dc.

My assumption (might be wrong or right) : In order to do an incremental differential backups of a volume, a full backup of this volume must be taken first as a base (only once is enough).

Current setup :
1) Volume share1 and share2 on Netapp NAS A are currently being backed up daily using differential incremental policies. On weekend/end of the month/year, we do full backups on these 2 shares.
2) All backups go to VTL (CDL A) pools and kept for 7 days before being vaulted to physical LTO tapes
3) All restoration is performed using the policy of NAS A.


Intended Plan :
1) First, one time full NDMP backups of volume share1 and share2 will be taken on NAS B (snapmirror/DR destination of NAS A). This full backup is stored on CDL B (same location as NAS B) and kept with a retention of infinity (or perhaps 6 months? I'm also considering doing full backup every 6 months instead of only once off)

 2) Once full backups  of share1 and share2 are completed, daily differential incremental backups of these shares will be backed up from NAS B for every weekdays. These backups will be stored on CDL B and kept with retention of one month.

3) On weekends/end of month/year, full backups of volume share1 and share2 will be taken from NAS A and kept on CDL A (retention time is not important here because we will vault these full backup images to physical tapes).

4) For restoration, if we need to restore from full backup, we will restore it from policy of NAS A, and for restoration coming from differential incrementials (daily backups), we will restore it from policy of NAS B.

Volume share1 and share2 on NAS A are being replicated to NAS B using snapmirror, so data on NAS A and B should be the same.

Thats pretty much it. Appreciate it if you can share your opinions whether this is a good implemention or what flaws or difficulties can you see from this plan, or even suggestions of better ways to do this.

Sorry if I'm being pushy..

I guess what I REALLY need to know is :

1) Is it OK to take NDMP full backup of a NAS volume and take NDMP differential backup of the same volume from another NAS (data on the 2 NAS are identical by way of replication - only 5 minutes delay between sync but I'm not too worried about that).

2) I assume that, before I take the differential backup from the other NAS, I have to do a full backup of the volume first as a base. After that, I will keep doing differential backup from this base. I will only do full backup once every 6 months on the other NAS to 'refresh' the base. Is there any problem that you can foresee with this?

If anyone can answer these 2 questions I would be really grateful...


Thanks!
Posie

as long as you remember that they are still essentially two separate hosts.

Personally, I wouldn't like the idea of trying to do a "full" restore, as it were, using the FULL from NAS A & subsequent differentials from NAS B - altho' I could well be wrong & there wouldn't be an issue. Maybe I just can't get my head around the logic (not yours!) of the state of the data at any one time.

***EDIT***
"Sorry if I'm being pushy.." - push as much as you want! Sometimes we see the post & think a while, sometimes we don't see the post (especially if there's been no replies as they don't stand out as much), sometimes we just don't know, sometimes........

Thanks for your reply Andy! :)

Yeah I still have some misgivings about backing up full from one host and incremental on the other (although data is supposedly identical if you ignore the 5 mins lag), but alas, we need to cut corners where we can and this is the best way I can think of...

Do you see any problem with taking a full backup only once every 6 months and doing incremental off that full backup every day? Since differential incremental should only back up changes that occur between today and the last backup, backup size wont be big (i assume). The only problem I see from this is that we need to restore a lot of incrementals  plus a full restore to get a complete restore. If you think of anything else, let me know?

- Posie

Thanks Andy :)
A bit of a late reply here, sorry. I had the same thought about the fulls's retention being long enough for the incrementals to be meaningful, but management here thinks that because our filer's restores are 99% not 'full restore' (in a sense that we dont really have a need to restore an entire volume fully. Their concern is more for the ability of individual files' restoration within that volume).

This is essentially how we got to the idea of performing full restore from one filer and the incrementals off the DR filer. In the event of disaster, we will only guarantee restoration from that full backup taken from the other filer (which we do every weekend and keep for a year). The incrementals we only keep to make sure we can restore individual files that change (rather than a part of a full, complete restore). So in this case, even if the fulls expire before the incrementals, these incrementals would still be useful. Do you agree?

 Anyway, you've been a lot of help, thanks a lot!

- Posie

As I'm in Holiday Mode & it's been a while can I just say "Yes" ?