Forum Discussion

milty99's avatar
milty99
Level 4
16 years ago
Solved

NetBackup DR Test

Hi

WE have a Netbackup Master Server at our DR site that backs up all of our servers at the production sites across a 10GB LAN extension. We backup to disk which stores data for 60 days and we duplicate to tape every day. We wish to use the NetBackup Master for DR tests as well as still use it to backup our live production servers. The DR servers will be on a different Network with different IP addresses from the production site but will have the same Node names.
Does anyone have any thoughts on how I can get this to work without building up a replica DR NBU server and copying the backup data on disk to the replica NBU environment?

Mark
7.1.x and Earlier
Backup and Recovery
NetBackup
  • John_Stockard's avatar
    John_Stockard
    16 years ago
    That sounds like a good solution.  Since your NetBackup master/media server and all of it's backup images are already at the DR site, I don't see a need to do any sort of NetApp SnapMirror replication (or any other replication technology).

    I still think that all you'd need to do in a DR situation to restore data to your DR servers is this:

    1. Have the DR hosting company provide you with replacement servers.  Make sure these replacement servers are connected to a "DR network" (the 172.16.x.x network listed above in my previous posts).
    2. Put operating systems on these servers (BMR could be an option for this, or you could use traditional methods such as Ghost or JumpStart or Ignite-UX or NIM depending on the OS).
    3. Connect your NetBackup master/media server to the "DR network" (172.16.x.x).
    4. Create hosts file entries on the NetBackup master/media server so that it can find the DR servers on the DR network (i.e., "zeppo-dr    172.16.20.4").
    5. Create a hosts file entry on each DR server so that it can find the NetBackup master/media server (i.e., "groucho   172.16.20.1")
    6. Install the NetBackup client software on each DR server.  This can be automatically added to the DR servers in step 2 as part of the Ghost/JumpStart/Ignite-UX/NIM process.
    7. Use the NBU Admin Console GUI to do redirected restores of the production server images to their DR equivalents (i.e., restore "zeppo" to "zeppo-dr").

    At this point, you should be back in business.

    You might want to seriously look at the BMR functionality included in NBU 6.0 and NBU 6.5 to help speed up step 2 above.  BMR used to be a separately-licensed product, but now it's automatically included with every "NetBackup Standard Client" and "NetBackup Enterprise Client" license.

10 Replies

Sort By
  • John_Stockard's avatar
    John_Stockard
    Level 5
    16 years ago
    One idea would be to create a bunch of entries in the hosts file on the NetBackup master server so that each DR version of a client has a "-dr" appended to it. 

    For example:

    Your production NetBackup master server is called "groucho" and has an IP address of 10.0.0.6.  You also have a production client called "zeppo" at IP address 10.0.0.12.  During your DR test you are going to be standing up a replacement "zeppo" server at IP address 172.16.20.25.
     
    In this scenario, you could create the a DNS entry on the master server that looks like this:

    zeppo-dr   172.16.20.25

    When you do your DR test, do a redirected restore of zeppo's backup images to zeppo-dr.

    Also, you will need to put a NIC into your NetBackup master server that can talk to the DR network (172.16.x.x in the above example).  The IP address on this NIC should also resolve to a "-dr" name (in this case if would be "groucho-dr").  You will probably also need to put a hosts file entry on each DR client that looks like this:

    groucho  172.16.20.6

    (substitute the real address of the NetBackup master server on the DR network)

    A possible downside of this approach is that you won't be doing any scheduled backups of your DR servers.  Your existing backup policies will still be pointing at the production (non "-dr") servers, not the DR replicas of the client servers.
  • milty99's avatar
    milty99
    Level 4
    16 years ago
    Hi John,

    Thanks for the post.

    Is this something you have tried? Would I need to create entries on the NBU configuration for each of the XXXX-dr entries?

    Would the DR clients require 2 NIC's one for the 10.XXX.XXX.XX address and one for the 172.XXX.XXX.XX ?

    thanks
    Mark.
  • zippy's avatar
    zippy
    Level 6
    16 years ago
    You said

     WE have a Netbackup Master Server at our DR site that backs up all of our servers at the production sites across a 10GB LAN extension.  (this one is used as youre DR master?)

    We backup to disk which stores data for 60 days (what are you backing up to disk? Catalog data?)

    and we duplicate to tape every day. (vaults right?)

    We wish to use the NetBackup Master for DR tests as well as still use it to backup our live production servers. ( you cannot do this and still backup your porduction network - Bad idea even if you can figure out how to do it, you could hose things up in production, I would seperate the two - DR in one subnet Production in the other, trust me.

    The DR servers will be on a different Network with different IP addresses from the production site but will have the same Node names. (install netbackup and restore catalog from a disk type catalog backup)

    Does anyone have any thoughts on how I can get this to work without building up a replica DR NBU server and copying the backup data on disk to the replica NBU environment?


  • John_Stockard's avatar
    John_Stockard
    Level 5
    16 years ago
    You would not need to create policies in NetBackup for the XXX-dr hostnames (unless you also want to do backups of the DR computers), but you will need to add entries to the hosts file on the NetBackup master server so that it can resolve the XXX-dr hostname to the correct IP address on the DR subnet (172.16.x.x in my above example).

    The DR computers would only need one NIC on the DR network (172.16.x.x), unless your DR test requires other connectivity (i.e., end-user testing).  The DR network NIC (172.16.x.x) is the only NIC needed in order for the XXX-dr computer to communicate with the NetBackup master/media server so that data can be restored to it.
  • milty99's avatar
    milty99
    Level 4
    16 years ago
    The Master server at our DR site is the only master server we have and is located at our DR site . We stage all data to disk (kept online for 60 days) and we then Vault to tape for long term storage. Catalog data is also put to disk then duplicated to tape. Is it is a bad idea keep DR and production running at the same time ? Would we need to maybe inline copy to another array and then mount the array on to a DR Master server to then perform DR tests. is that the best option? We are trying to avoid recovering from Tape when we have a perfctly good disk array with all of the data required for DR in the same location.
  • Alex_Vasquez's avatar
    Alex_Vasquez
    Level 6
    16 years ago
    In a DR scenario the thought is that your primary production site goes down and is inaccessible.  To that end how can you use your existing disk to recover to your DR site?  Well, you're going to need a DR server configured similarly to your existing prod. server.  If you have a SAN at both your DR and prod site, you could put your catalog files on a SAN attached LUN and then replicate that LUN to your DR site and present the replicated LUN to your DR server.  That way your DR server would have a catalog available to use and you could forgo using tape.  Then any recovery you needed to do would be done from your archived tapes.... 

    If that's not an option, using a SAN replication technology via EMC or Netapp, then you could use Vault or SLP, I'd imagine, to dupe the data to DR.
  • milty99's avatar
    milty99
    Level 4
    16 years ago
    Hi Alex,

    All of our data is backed up across a 10GBit IP connection to the DR site which has our NBU environment. The DR site is at a hosted DR facility.  When we invoke DR or require to run DR tests they will provide us with DR servers. The RPO and RTO required by the business does not warrant spending on replicated SAN infrastructure ( we have Netapp).
  • John_Stockard's avatar
    John_Stockard
    Level 5
    16 years ago
    That sounds like a good solution.  Since your NetBackup master/media server and all of it's backup images are already at the DR site, I don't see a need to do any sort of NetApp SnapMirror replication (or any other replication technology).

    I still think that all you'd need to do in a DR situation to restore data to your DR servers is this:

    1. Have the DR hosting company provide you with replacement servers.  Make sure these replacement servers are connected to a "DR network" (the 172.16.x.x network listed above in my previous posts).
    2. Put operating systems on these servers (BMR could be an option for this, or you could use traditional methods such as Ghost or JumpStart or Ignite-UX or NIM depending on the OS).
    3. Connect your NetBackup master/media server to the "DR network" (172.16.x.x).
    4. Create hosts file entries on the NetBackup master/media server so that it can find the DR servers on the DR network (i.e., "zeppo-dr    172.16.20.4").
    5. Create a hosts file entry on each DR server so that it can find the NetBackup master/media server (i.e., "groucho   172.16.20.1")
    6. Install the NetBackup client software on each DR server.  This can be automatically added to the DR servers in step 2 as part of the Ghost/JumpStart/Ignite-UX/NIM process.
    7. Use the NBU Admin Console GUI to do redirected restores of the production server images to their DR equivalents (i.e., restore "zeppo" to "zeppo-dr").

    At this point, you should be back in business.

    You might want to seriously look at the BMR functionality included in NBU 6.0 and NBU 6.5 to help speed up step 2 above.  BMR used to be a separately-licensed product, but now it's automatically included with every "NetBackup Standard Client" and "NetBackup Enterprise Client" license.
  • Alex_Vasquez's avatar
    Alex_Vasquez
    Level 6
    16 years ago
    John's give some excellent suggestions, be sure to give him points for his efforts! =)  Otherwise the Stumpr will come and get you!!
  • milty99's avatar
    milty99
    Level 4
    16 years ago

    I have tried the above and it works very well. The only issue I can see with the solution is the problem of potentially (finger trouble/ lack of concentartion under dr invocation/late nights etc) where the restore is directed to the production system rather than the dr system. We will be looking closely at how we can prevent this from happening.