I don't see in here how much data you're talking about, but another option is 
to split your problem up between uptime and data protection...

If you lose a server at a remote site (multi-drive failure in a raidset, 
spilled coffee, whatever) I can't see making that 4-hour SLA unless you have 
another server waiting to go.  I would explore a 3-layer protection approach: 

layer 1-  The Oops factor: snapshots. Users recover their own deleted/changed 
files.  You're out of the loop and they control the restores (figure you set it 
as many times a day as you want, set aside disk).  Most restore requests of 
this type are in the first couple weeks anyway. (MOST- not ALL)

layer  2- Delayed oops factor: remote protection: put a Puredisk de-dupe 
solution in place to push the data back to your central location.  Restores 
that have rolled of the back of the snapshots above are recoverable from 
central storage.  Takes a bit longer, but often these are less pressing.

Layer 3- Availability- Have 2 file servers at each site with a real-time 
replication between them (ie: Repliweb, DFS-R or some such).  Use DFS- one 
server dies, re-point the DFS and you're back in business.  Less than 4 hours 
(maybe 4 minutes!) and you're on line.  Make sure your 2nd server has snapshots 
and remote replication going too, so you keep the history.

Sounds expensive, but look at what you get: No remote tape, no remote backup 
server, fast recovery (2 ways) and long-term offsite storage.   You're are 
already running 100+ tape libraries and backup servers- why make users wait for 
a restore from tape when you have a perfectly good server sitting just inches 
away?

Tom


srabbi wrote:
> Hi All,
> 
> I am designing a centralized backup/restore solution for 100+ branches. We 
> are currently using NetBackup to do the data backup at a local tape 
> library attached to each branch server. Our business folks do not want to 
> mount/unmount tapes for weekly and incremental backup. So my task is 
> to find an innovative way to solve this tape handling problem. We also have a 
> very strict SLA for data restore ( 4 hrs) with business. 
> 
> I am wondering if I can keep the local tape library in a lock down state and 
> use it for weekly full backup (i.e overwriting tapes) and do the incremental 
> back at the central site. This means that each NetBackup client needs to talk 
> to a Master and Media Server for the Full backup and then talk to another 
> Master/Media Server for incremental backup over the WAN.  Can a NetBackup 
> client configured to talk ot multiple Master server?  IF so, what is the way 
> to configure it? 
> 
> My Branch servers are all Win2003 Servers and have medium speed WAN link. Any 
> insight to the proposed solution would be much appreciated. I exhuasted 
> seraching NetBackup documentation and I could not find an answer (yes/No) to 
> my question. I hope Fourm members can anser to my question based on their 
> experience. Thanks


+----------------------------------------------------------------------
|This was sent by tburrell_mn AT yahoo DOT com via Backup Central.
|Forward SPAM to abuse AT backupcentral DOT com.
+----------------------------------------------------------------------


_______________________________________________
Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu