Backup Index Replication

[teaboy]

I am in the process of planning a hot (or at least warm) backup site for my data center, and I'm trying to determine the best way to have restores available in the quickest manner.
The site will have similar servers and tape library to our current configuration.
I was wondering if there is any way with Networker 5.1 (yes, that old) to replicate the backup indexes to my hot site. This would elminate, or at least substantially decrease, the need to read bootstrap/tape information in order to get the tape into the media db and indexes. I would think this would lower the time needed to be ready to restore, but maybe I'm wrong.
Could someone please tell me if you've had experience with such situations/planning, and whether I'm barking up the wrong tree or not?

Thanks

T

 

[cptb]

To get the best restore performance for a NetWorker client data the data needs to be read from tape in one sequence. This usually means that you need to de-multiplex the tape and write it to another one. The way to do this is to use saveset cloning. Cloning however, can be a slow process, depending on how many savesets you have multiplexed on a tape volume, so the fastest way to clone would be to read the data from a diskdevice. If you want to secure your NW server you can ofcourse use a replication software to replicate it another site. Thats not really a supported solution from Legato and you need to make sure you don't have the same licenses on both machines. A supported solution however is a cluster solution with MSCS or Legato Cluster.

 

[teaboy]

Maybe I'm missing something - what "data" are you talking about writing to disk?
I'm not concerned so much with the speed of restoring information from tape vs. restoring it another way, I'm more worried about the time of getting the tape volume information into the server in order to have the information/tape indexed. It's been my experience with doing things similar in the past that importing tape information into an index in order to start restoring from the same tape(s) is a painfully slow process. This is why I was seeing if there's a way to replicate the index/media databases, not the data itself. The scenario would be:

1. Current building blows up/burns down.
2. Fly to new site with most recent backup tapes retrieved from offsite storage.
3. Arrive at hot/warm backup site.

At this point is where having updated indexes could make a difference. The choices would be:

4. Insert tapes into library, inventory, and start restoring.

OR

4. Insert tapes into library, inventory, find and restore indexes, start restoring data.

I would think the second option would be much more time consuming and labor-intensive than the first.
Again, I assume I'm looking at this in the right way. I've looked through their disaster recovery guide, and it seems to follow what they have in there.

 

[cptb]

Hmm, Ok.

Running scanner to rebuild the indcies can be a time consuming process yes. An option to restore data to a networker client can be to use SaveSet recover instaed of index based restore. But in the long run you're right about a replication solution being the solution to go for. Legato have a replication product for the Windows platform that can be used. It is not a syncronous solution as their old Unix product but it could be an option anyway. Or, take a look at the Recovery manager. It currently supports Solaris and Windows.

 

[Guest_imported]

No, No, No....
I use mmrecov to rebuild all the indexes from the bootstrap information.
Below is the Doc I wrote for myself to use in a firedrill.
============================================================
Recover Index using Bootstrap Info


Get save set ID using 1 of 2 methods.
1. Retreive from printout of bootstrap info
2. Put most recent tape into drive, and type scanner -B /dev/rmt/0hbn

Note the Save Set ID, and Volume Label of desired backup

type mmrecov
Answer save set ID, starting file number, and recover number to start

After Index retrieval, restart networker

============================================================
It fulls our complete index off of tape in under 5 minutes from anywhere on the tape.

 

[cptb]

Well, you need to keep things apart. mmrecov will rebuild the BU server index only. Of course from that you can do a regular restore to restore client indexes. That is a timeconsuming process to. With the replication scenario you wont have to restore any indexes.