Hi
We have this kind of scenario running.
However, we've used HACMP 4.4 in a active/passive mode to be able to use
the fail-over scenario, and are only running one active instance of TSM on
the cluster. There is one active and one passive clusternode. The config
is a follows:
- 450 clients, divided into 250 UNIX servers and 200 Windows servers.
- Two clusternodes, based on the P-Series 660 with dual drawers, 6
processors, 8 HBA:s and 4GB of RAM
- 3 active cluster LAN connections, and one heartbeat connection. Also,
each cluster node has a private connection.
- SAN-based storage devices. For primary stgpool they use 9310 with 12
9840B drives, and for the disasterpool they use a 3494 with 4 3590E
drives.
- Approx. 3-4TB of data transfer each night.
- A total of 90TB in primary pools, and 90TB in the copypool.
The reason we only use one active instance, is cause the customer wanted
to have minimal administration. When using two active instances, you'll
have to administer two TSM servers. Therefore, we sized the TSM servers
accordingly.
Also, now the customer has the possibility to do HW and SW upgrades,
without interrupting normal operations. He can just add more HW, or
upgrade the SW, on the passive node, and then do a cluster rg move to this
node, and then do the same changes on the other node. The customer also
has the possiblity to test a software upgrade before commiting it, by
upgrading the software on the passive node, test it, and then decide
wether or not to commit the upgrade.
If the primary site goes down, the passive node will go up as active.
Restores will be possible, as TSM will do the restores from the copypool,
if the primary pools aren't available.
Best Regards
Daniel Sparrman
-----------------------------------
Daniel Sparrman
Exist i Stockholm AB
Propellervägen 6B
183 62 TÄBY
Växel: 08 - 754 98 00
Mobil: 070 - 399 27 51
Koen Willems <koen_willems AT HOTMAIL DOT COM>
Sent by: "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
2003-03-24 11:27
Please respond to "ADSM: Dist Stor Manager"
To: ADSM-L AT VM.MARIST DOT EDU
cc:
Subject: hot standby cross failover tsm servers
Dear listers,
A customer wants to build a TSM config consiting of two tsm servers on AIX
both in a seperate datacentre with a 2 mb framerelay connection in
between.
They use fibre connected 3584 libraries on both of the sites.
The customer wants to place his drm volumes as hot standby in the library
within the other datacentre.
He would like to sync his TSM databases on both site so when a disaster
strikes the can start restoring with no extra effort in the other
datacentre.
My plan was to split both 3584 in to two logical libraries and build a
second tsm instance on both pSeries. When the DRM volumes arive at one
site
a script does a dsmserv restore from the DRM database and in this way the
drm volumes are availleble at the hot site.
Is there a way within TSM or on a lower level to just sinc databases
remote
via framerelay ?
In this way I dont have to work with a dsmserv restore routines...
Best regards,
Koen Willems
tectrade bv
The Netherlands
_________________________________________________________________
The new MSN 8: advanced junk mail protection and 2 months FREE*
http://join.msn.com/?page=features/junkmail
|