08-10-2012 12:52 PM
Question #1
what could be best and cheep solution in the following scenerio:
What least fit in my mind is SFHA at primary site and SF at secondary site. Primary site has VCS installed and configured and replicating data at secondary site with VVR. So the company only need to buy:
Because of this company can get HA for primary site and can secure its data via replication/vvr at secondary site.
Question#2
Is there any thing in veritas SFHA regarding standard license and enterprise license. If any details please share
Solved! Go to Solution.
08-13-2012 01:11 AM
What you propose will work - but more info on your comments:
Enterprise license and GCO are 2 different licences - you do not get GCO with Enterprise license and I THINK Enterprise license is a lot cheaper than GCO licence. Enterprise license gives you things like fast resync mirrors and space optimised snapshots.
With campus cluster and stretch cluster you can actually have manual failover (you set service group AutoFailOver attribute to 2), but this requires the same license to GCO (HA/DR license).
What I meant by "some put a one-node VCS cluster at secondary site without GCO" is a separate one-cluster is configured at secondary site, but GCO is not used to connect the 2 clusters. This just means you can mount your filesystems, start your applications, etc, at secondary site with VCS in one step, rather than having to start all these components manually with several steps.
Regardless whether you put VCS on secondary node or not, if you do not use GCO, then I would not use Replication agents as without GCO, you have no coordination between sites and therefore performing a VVR migrate or takeover should be a manual step to prevent split-brain (and it may be that using replication agents requires HA/DR license anyway).
Mike
08-11-2012 03:01 PM
Proposed solution is ok and I have seen a few customers do this (some put a one-code VCS cluster at secondary site without GCO). Enterprise license is not required - this is mainly for fast resync mirrors which you don't need if you are using VVR, although Enterprise licence also gives you space optimised snapshots which means you can bring up secondary site on a Copy-On-Write snapshot which typically only requires 10% more storage for snapshot (this feature is called Firedrill).
Mike
08-12-2012 09:29 PM
Thanks for your kind words Mike
Enterprise license means you are talking about GCO ? If yes then it involves cost thats why its excluding from the proposal.
You may talking about the campus cluster/Stretch Cluster. This required one cluster for both sites. and data is replicating via mirroring at Secondary Site ? If yes then the problem with this is actually auto failover. which you cant make manual from primary site to secondary site. If need manual failover between sites then should buy the GCO option(which you are say "Enterprise license" ) which itself a big cost actually.
====================
So this will work....
Primary Site Secondary Site
SFHA installed SF installed
Two Node Cluster configured
Node-1 Node-2 ======================================Node-3
VVR is replicating data
between
Primary and Secondary SIte
08-13-2012 01:11 AM
What you propose will work - but more info on your comments:
Enterprise license and GCO are 2 different licences - you do not get GCO with Enterprise license and I THINK Enterprise license is a lot cheaper than GCO licence. Enterprise license gives you things like fast resync mirrors and space optimised snapshots.
With campus cluster and stretch cluster you can actually have manual failover (you set service group AutoFailOver attribute to 2), but this requires the same license to GCO (HA/DR license).
What I meant by "some put a one-node VCS cluster at secondary site without GCO" is a separate one-cluster is configured at secondary site, but GCO is not used to connect the 2 clusters. This just means you can mount your filesystems, start your applications, etc, at secondary site with VCS in one step, rather than having to start all these components manually with several steps.
Regardless whether you put VCS on secondary node or not, if you do not use GCO, then I would not use Replication agents as without GCO, you have no coordination between sites and therefore performing a VVR migrate or takeover should be a manual step to prevent split-brain (and it may be that using replication agents requires HA/DR license anyway).
Mike
08-13-2012 02:52 AM
therefore performing a VVR migrate or takeover should be a manual step to prevent split-brain
Definately as an example in the Migrate Scenario : we have to stop Application Service Group resources only (Not Replication Service Group resources) , then do migrate , then up the Application Service Group resources at secondary site.
Thanks Mike for your kind and superb words on the query. Marking the post as solution
(other ppl can give there comments after mark the post solved)