Let us take a look as to how an Outlook client connects to an Exchange 2013 mailbox.
Outlook queries Active Directory for SCP record, which is the AutoDiscoverServiceInternalURI set by the Exchange administrator. Active Directory hands back the autodiscover url and the Outlook client connects to the autodiscover service and gets the information for all the Exchange urls.
Assuming that the CAS servers are load balanced using NLB or HLB, Outlook connects to one of the CAS servers. The following happens after that. (The below diagram is from one of the TechED NA Exchange session).
- CAS servers authenticates the incoming connection.
- Once the user is authenticated, CAS looks up where the user mailbox is.
- CAS server finds which mailbox database hosts the mailbox for the user and in turn finds which server hosts the active copy of that mailbox database by querying the Active Manager.
- CAS proxies the connection back to the mailbox server which is hosting the active copy of that mailbox database.
- For any mailbox connectivity, the user is always served by the server that hosts the active database copy. CAS only authenticates and proxies the connection.
- Within Outlook, the server name will be the mailbox guid @ the smtp domain. This is different to an Exchange 2010 connection where the server name is the CAS array.
- Next time a connection is initiated by the same user, any one of the CAS server will pickup the connection and proxies it back to the relevant mailbox server. As all CAS servers will have the same certificate, the user will not be prompted again for credentials.