soliciting feedback on the proposed DR configuration for netbackup....
existing environment: 1 storage domain 1. Netbackup Master (Intel) and Media Servers (Intel and Unix) 2. IBM LTO 3584 Tape Library (connected via fiber)
proposed DR environment - configure a new storage domain with the following components: 1. Second Netbackup Master Server (Intel) 2. IBM LTO 3582 Table Library (connected via fiber)
Objective is to be able to restore files in the event of a disaster at the production site and minimizing the time required to re-install the DR backup infrastrucure. In addition, please advice if its possible to replicate the Image catalogs from the production master server to the DR master server to allow immediate restoration of files without having to rebuild the catalogs from tape.
IF you have SAN Attached disk on both Prod and DR Master servers you can setup a nightly replication of the database to your DR side thus allowing you to bring up once the link is broken between the two sites and this should allow you to be up and running shortly without doing any restores or rebuilds. If you want in-depth details on how to accomplish let me know and I'll give you info on how we recover. Thanks.
Thank you very very much !!! As you mentioned, would very much appreciate to explore probable/feasible solutions for our DR netbackup infrastructure more in detail......
In addition to what was mentioned...... the proposed DR infrastructure (second Master Server with its own resources) would not be connected to a SAN...thus, if we do implement replication of image catalogs from the production master server, it would be from production SAN (attached to the production Master Server) to the DR Master Server (locally attached storage). Having said that, would our proposed solution be the most ideal solution...considering we would like to have the DR infrastructure be readily available in the event of a disaster (no rebuilds required)...
Ken, Would the Dr server be attached to a Dr SAN or is it just using locally attached disk? Are the rest of your DR servers just using local disk also or do you have a separate DR SAN. The way I have it is a cheaper disk array located at my DRsite that My Netbackup and all other servers are attached to, that way I can replicate all data including the Netbackup catalog out to the site each night. If you are unable to do that your solution would be as you described, you would have to back up your catalog to tape and take that with you in the event of a disaster, restore the catalog to the local disk and go from there.
The way we do it, you do not even need a backup of the catalog on tape. On our production master server, we take a backup to disk of the catalog and immediately after, we replicate the disk backup (using Veritas Repliation Exec) to the DR master server. We also replicate the images directly to this DR server every 2 hours.
This way, the steps to activate the DR server: - change the NetBIOS name of the DR server to the original master server name - install netbackup - import the catalog (bprecover -r -dpath)
Ken, I would also suggest exploring a VTL (Virtual Tape Library). With this, you do not need a SAN at the DR site, you can simply replicate virtual tape images to the VTL at the DR site. From there you can rebuild the catalogs. A VTL emulates a tape library either using SAN or direct attached storage. In your case it would be direct attached storage.
I have a very similar situation with one twist. My DR Master is backing up servers located at the DR site. I am currently replicating the Image folder to the DR Master in prep of a disaster fail over. In the event of a disaster, it seems like my only option would be to overwrite the catalog at the DR site. What additional steps will I need to do now to prepare for a graceful failover without losing the catalog at the DR site.
I read that for NBU6, one can recover only the catalog image file to alternate server. Maybe this feature would helps.
For NBU5, which is my current setup, I've no choice but to configure at DR site a media server (communicating to master server at production site) and meanwhile having a cold standby master server at DR site.
During DR, the cold standby server would boot up as master server and the media server at DR site could configured (by changing the IP in hosts file) to point to this master server at DR site. This approach preserve the backup images done at both production & DR sites.
I've pretty much done the same thing and configured it as someone suggested earlier. I do a full backup of the catalog to disk and that gets replicated to the DR site. I'm still waiting to upgrade to 6.x and hopefully the new version will make a lot of this easier.
I'm currently looking at our options for replicating the DR catalog over from our main site to DR site at Sungard. I would be extremely greatful if you would you be kind enough to provide the in-depth details on how you have accomplished it/how you recover as mentioned in your post.
My work mail address is mike dot haslam at tnt dot com