Enterprise Vault Building Blocks İnformation

[vaultlearner]

Hello,

I need get more information about building blocks. As I understand from Admin Guide. I need SQL server separate and all partitions on shared storage. And all Extra Requirements.

Lets say I am archiving Journal and Mailbox. I installed already EV and I want to use Building Blocks. Should I add new idle Enterprise Vault server to existing Directory when I installing it?

I want to install Idle EV server same version as Node1 if something goes bad for Node1 just want to pass to node2 by changing Alias from DNS.

What is the recomended Installations for Building Blocks.

Thanks

VaultLearner

[GertjanA · Accepted solution]

Hello again.

I can explain, but I seriously advise to discuss in person with someone who knows (support? Consultant?) to make sure you got everything covered.

Scenario 1 - I assume you only have 1 EV server. I assume you have 1 SQL server, I assume you have you indexes/data local to EV server. SQL HA/Clustering/Mirroring/TransActionlogshipping for HA of EV databases. install 2nd EV server, in same site, new alias. no need to install services etc. Make sure you somehow replicate te data/indexes to this 2nd server (make sure to use same paths/locations). Failover = stop all EV services, except Admin and Directory. update DNS (i.e. switch aliasses. run final sync on data/indexes to make sure both are identical. in EV, run USL (Update Service Locations), only after you verified the aliasses resolve to the changed names (ping aliasses from both servers). USL will make sure functionality is switched (non-existing services will be installed, tasks will be moved etc). When done, start other EV services. done. (only EV failed over). For SQL make sure you also have an alias for that. if you need to switch sql also, update SQL alias to point to new SQL server, restart ev services, off you go. note that indexing might take a while to start, as it has to upate/create the indexmetadata on the 2nd server.

Scenario 2 - I assume SQL is clustered, data and indexes are on SAN. install 2nd server as above, same procedure, make sure data and indexes are accessible from 2nd EV server.

If you use SMTP archiving, there are additional things you need to replicate. see admin guide

Also see NO1

in general, folders you have defined on ev1 need to be also available on ev2 (think VSA temp folder, shopping folder, storage queue etc.) BEfore performing this (in a 'test', set the archiving tasks in report mode, restart them, wait until both MSMQ and Storage queue are empty. Then set backup mode on indexes and vault stores. then stop the services as indicate in scenario1, replicate data, perform failover. start ev2, verify no events are shown related to accesssibility of databases/data/indexes. the clear backup mode, verify again events. then start the task in normal mode.

[vaultlearner]

Hello GertjanA,

Thank you very much for detailed information :) 

I understand senario 1 and 2, when we add new Ev server in the same site normally my senario was using it idle. But as I understand we can use it for archiving purposes. For example we can use ev1 for journalling and ev2 for email archiving when the time comes to switch it we can easly switch the healty one. Of course every you said must be granted like storage, folders etc.

Thank you again explanation. And I am going to try this on my lab first.

Vaultlearner

[GertjanA]

Hello, I am glad it was clear. You're assumption is correct.

The way I have it is (only Journal Archiving) 6 Journal Archiving servers in DC1, 6 in DC2. Both archive from an Exchange Journal DAG, consisting of 12 servers. In a DR scenario, we'll change aliasses of EV servers in DC1 to point to DC2, then update service locations. That way, the funcitonality of the EV Servers in DC1 switches to DC2 = 6 servers performing the work of 12. For these 12, we use an index server group, consisting of 2 indexing only servers in each DC. In a failover, we also switch DC1 index to DC2 index = 2 indexing servers doing the work of 4. Some backlog will occurr, but that is an accepted risk...

In addition, we run EV SMTP archiving also, these servers are VM's, and we use SRM to cover fail-over. No backlog whatsoever. The EV SMTP servers run indexing local, which makes it easier to failover. Only thing we need to adjust for the SMTP part is update the loadbalancer, to point to the new ip-addresses.

Once setup, it is really error-free. Planning is key.

[vaultlearner]

Hello,

Thank you for the real life example. I'll test and get to know better on lab.

Thanks,

Vaultlearner.