Anybody ditched tapes completely/cheap array recommendations?

datsunman

datsunman

datsunman

Associate
Joined
6 Aug 2003
Posts
336
Thinking through the logistics of dropping tape backup and going completely D2D. Anybody done this?

I was thinking along the lines of two large SATA arrays, one in prod site one in DR. D2D in prod and then replicate over to DR. Replace the drives/array every three years.

Initial thoughts suggest there would be reasonable savings.

Thoughts?

Looked at Nexsan, any other recommendations for a large SATA array?
 
IMO D2D still has gaping holes for anyone smaller than enterprise with lots of sites (more than 3 anyway). For smaller organisations, tape gives enough flexibility to workaround many issues.

- How will you deal with long-term archives?
- If the DR building is on the same site as primary then could it be affected by the same risks at the same time i.e. fire, flood etc taking down everything.
- If DR is located on a different site are your redundant data links fast enough to still allow replication? The primary link may be e.g. 1Gbps but what happens if that's offline and you have to use 15Mbps failover for 10 days?
- How will you maintain offsite replication capability if the DR building/array gets damaged?
 
Yes, we have dumped tape.

We have 2 SAN's. One here and one at our DR site. These replicate differences every hour.

Once its working. Its fantastic.
 
We've been trying to do this for a while now - disk to disk (with de-duplication) then the backup is copied within the backup system to a different site, but its not quite got there yet due to some backup software 'development issues'. Intention is to retain the tape library only for monthly copies/archives.

I'll be glad to ditch the daily tape run!
 
Its been bought in but we were pretty early adopters of the vendor's de-duplication software....not quite the polished product that was expected but its getting there now.
 
Hairy said:
Yes, we have dumped tape.

We have 2 SAN's. One here and one at our DR site. These replicate differences every hour.

Once its working. Its fantastic.
Click to expand...

There's a difference between this and backup though... If all data on SAN1 corrupts, SAN2, within an hour, is filled with corrupt data. Then what?
 
We have a DR mirror of our SQL,Exchange,Fileshare.
Everything is backed up to disk during the week and backup to tape at weekend.
 
oddjob62 said:
We have a DR mirror of our SQL,Exchange,Fileshare.
Everything is backed up to disk during the week and backup to tape at weekend.
Click to expand...

I think this is the ideal situation.

We have a similar system - although we don't have a DR of our SAN.

However we backup to disk nightly and then to tape every weekend. These are archived for a month offsite.
 
brainchylde said:
I think this is the ideal situation.
Click to expand...

Yeah it seems to work nicely for us (using MS DPM if anyone is interested).

1. Any regularly changed data is replicated offsite (exchange CCR geocluster, SQL mirror and file DFS).

2. There are also regular back up disk (exchange/sql every 1h, fileshare 3x daily)

3. Any servers that might be needed in a DR scenario are backed up to DR site.

4. Everything is backed up to tape weekly and sent offsite (ironmountain)
 
It very hard to ditch tape as there are a lot of regulatory and statutory requirements for the retention of data. When this sort of thing is required the we usually design in a D2D for performance (snapshot and replication if required) and then a dump to tape to limit the impact of the slow tape drives on the live systems.
 
brainchylde said:
There's a difference between this and backup though... If all data on SAN1 corrupts, SAN2, within an hour, is filled with corrupt data. Then what?
Click to expand...


Not quite,

It replicated changes in data as separate files.

You you can track back to a 'Backup' before the corruptions occurred.
 
oddjob62 said:
Yeah it seems to work nicely for us (using MS DPM if anyone is interested).

1. Any regularly changed data is replicated offsite (exchange CCR geocluster, SQL mirror and file DFS).

2. There are also regular back up disk (exchange/sql every 1h, fileshare 3x daily)

3. Any servers that might be needed in a DR scenario are backed up to DR site.

4. Everything is backed up to tape weekly and sent offsite (ironmountain)
Click to expand...

This does seem like the perfect scenario.

I'm sure its very expensive getting fully mirrored kit!
 
Hairy said:
This does seem like the perfect scenario.

I'm sure its very expensive getting fully mirrored kit!
Click to expand...

Quite - having said that, your backup/replication plans should be dictated by business requirements rather than ideal scenarios. If you can afford fully replicated infrastructure I guess the demands placed upon your infrastructure really require it. And vice versa.

Actual backup solutions aren't that expensive to implement. Storage is cheap and backup software isn't massively expensive. There is certainly no excuse for anyone not to backup. :)
 
Some food for thought there.

Our backups are sent to disk every night and then copied off to tape during the day. All servers are replicated to DR, either on continuous or nightly schedule depending on content.

I have very low storage requirements so replicating the disk backups (and storing multiple copies) would not pose a significant problem.

J1nxy - any examples of regulatory/statutory requirements? I've been looking for written examples but struggling to find anything.
 
You must log in or register to reply here.
Back
Top Bottom