EV10 storage

[Sortid]

Reading the release notes for EV10, can someone tell me exactly what

Indexing and storage no longer need to be co-located

means pls and how that relates to previous versions?  I thought Indexes could always be stored away from storage...

[JesusWept3]

It means that you can have a vault store on one server and indexing on another server.
In EV9 and below an indexing service HAS to reside on the same server as a vault store

Say you have two EV Servers
EVServer1 and EVServer2

Lets say you only have one vault store on EVServer1 and EVServer2 is idling.
You might want to put Storage on EVServer1 and then have EVServer2 do indexing, in EV9 this is not possible, the indexing service has to be on the same server as the storage it needs to index.

The idea is in EV10 you could have
EVServer1, EVServer2, EVServer3, EVServer4 to be storage and archiving and then have
EVServer5, EVServer6, EVServer7 and EVServer8 be dedicated to indexing, meaning that you could have Discovery Accelerator searches run a lot quicker and not impact any performance on the Storage servers

[MichelZ]

Yes, they can be stored "away" from storage, but as JesusWept2 explains, the "Processing" needed to be on the same Server.

Now you can get more servers for Processing the Index (very CPU intensive), so you are able to scale better, and have the actual Data (Vault Store) just on one Server.

Cheers

[Sortid]

So how would you determine if you either need a separate Index server or multiple index servers?  Our org has 7000 users, and about 4 people using DA, so while I don't think it gets hit that hard, what are the signs that this should be looked at?  Slow searches?  Is there a sizing calculator?

[JesusWept3 · Accepted solution]

I don't think theres a sizing calculator, but its not really the amount of users you but how much you archive, i've seen companies with 300 users dwarf the amount of email sent/received by an organization with 5x those numbers.

But really if you see that Searches are taking longer and longer or taking up a lot of resources, it might be a good idea to maybe have 4 storage and 2 index servers, but if it really isn't broke then don't fix it.

Another advantage is of having a seprate index server is that you can free up resources for your storage servers in terms of available disks and disk space and you're not having any contention of indexing and storage fighting for resources etc

[AndrewB]

sortid, since most of the environments i work in are using a combination of SAN for the indexes and NAS for the vault stores, i really see the advantage of seperating the storage and index servers when it comes to cpu and memory. i envision the adoption of this concept to be fairly limted and to mainly apply to large enterprise environments that are journaling massive amounts of messages each day and where constant discovery of up to the minute data is a requirement.