LTM Sizing for Exchange 2010

Casey,

 

 

What is the primary access method for your Exchange environments(OWA, ActiveSync, RPC, RPCoverHTTP, etc.), and what is the concurrency of the connections? The number of mailboxes hosted is not directly correlated to the F5 device needed to support the users.

 

 

For example, two customers can have the same number of mailboxes hosted - let's say 130,000 - but one expects 10% concurrency vs another expects 50% concurrency.

 

 

Furthermore, based upon which access method is used for Exchange, the overall traffic generated by the client is different - OWA interface, for instance, is MUCH richer in Exchange 2010 than it is in 2003 or even 2007 - so from pure HTTP perspective, a single OWA user is going to generate a lot more traffic and requests/sec using OWA 2010 than they did with older versions.

 

 

Even for RPC-only deployments in Exchange 2010, all traffic is nowgoing to CAS array, vs in 2003 and 2007 RPC traffic was going directly to the mailbox server - and since F5 is load-balancing CAS array, it means that all RPC traffic is going to go through F5.

 

 

Sitting down with your local F5 team to best understand the requirements and nuances is the best recommended approach - and you are always welcome to post specific questions on this forum as well.