HA/Failover with EV

[Robert_Ton]

I am looking into a HA solution in my EV environment. The two options I know of are:

1.) Building Blocks solution using Service Locations
2.) VCS with EV SP2

Our SLA's do not require as quick a failover as VCS provides, but we do not want to jump through a bunch of manual steps either.

Does anyone have any experience with any of these options?
Are there any manual aspects of VCS or is it completely automated?

TIA

Bob

[Rick_Krieger_-_]

The USL failover option is actually very quick and simple (i.e. approx. 1 page of steps in admin guide... pg 201.) The main difference between SFW-HA (VCS) and USL options is automated vs. non-automated.

IMHO, two key deciding factors wouId be considering how you plan on making the data available on the standby node (Vault stores, indexes, shopping dir, etc.) and... whether or not the mechanism for initiating failover needs to be manual or automatic.

If you have a simple, reliable, repeatable method for making this data visable on the standby server and your SLA will allow you to initiate failover manually then then USL could work nicely for your situation. (you could also implement SFW to help with the making the data available on the standby server and still just use USL to do the failover)

If you have more involved data visibility/data replication needs and/or your SLA requires automated failover then you may be better served using SFW and/or VVR for data presentation and SFW-HA / cluster option with EV 6.0 SP2 for the automated failover piece.

Hope this helps,
Rick

[Askar_Kopbayev]

Are there more detailed information on building and administering both solutions?

[Robert_Ton]

Rick - Yeah, I saw the page of steps in the admin guide, but I was suspicious if it was as easy as they diagram. Do you have any experience failing over EV using USL?

Also they mention a working building block's solution is required, but nowhere can I find out how to build out such a solution.

Thanks.

[Rick_Krieger_-_]

:) Yeah... working building blocks solution basically means... EV configured, and working, components where they need to be to allow for failover in logical chunks, etc..
re: USL... yes, I've done it quite a bit and it is a piece of cake.
I recently had a situation similar to yours...
i.e. existing EV server, working but OS hosed and needed to be rebuilt, built new EV server, used USL to failover to new server, fdisk'd existing server and rebuilt, then did a 'recovery' back to original server. no problemo.

the only thing you may need to spend time on working out is... how to make the vault store data, indexes and shopping data appear on the new server.... if you are using SAN LUNs presented to both hosts (like you would do for a cluster) then this is not a big deal. If not you either need to 'bolt' the disks/data into the new server or you could use a data migration tool to copy/migrate the data over to the new server (it's just a bunch of files in after all.) You just need to make sure the EV services are turned off when you make the final sync if you are going the copy/migrate route.
Bottom line... you need to make the vault store data, indexes and shopping data appear on the secondary/failover server so when you do the failover and the services start up EV can see the data.

SQL is easy in your case since it is on another server.