I am currently in the process of testing Exchange 2010, and although I have my head wrapped around DAGs, I have some questions regarding CAS functionality. We are going to be deploying mailbox servers among two sites. Each site will have active and passive mailboxes (two DAG groups total), and each site will have multiple CAS servers. My question is this. I want to provide a single URL for users to login to OWA, and for Outlook auto configuration. But if I do that, how does Exchange know which site a user's mailbox is in? I don't want them logging into OWA on Site A, if their mailbox is in site B, because it's just going to be extra overhead on our point to point circuit. I also don't want to give some people one URL, and some people another. If Exchange handles this automatically, great. If not, is there something I need to architect here to get this to work the way I want?

So you have a couple of options, but I would strongly recommend against trying to get a single Outlook Endpoint FQDN across both sites. This will causes issues for you and will make failovers etc. a lot more complicated and increase WAN utilization. So create a CAS Array for each site with a unique FQDN for that site. The RPCClientAccess property of the database will then point to the CAS Array in which the database will be normally active. If the DB goes active in the other site, the CAS Servers will RPC over the the mailbox server directly, so you need to keep track of what database is active where. In order to achieve what you want for Outlook Web App you are going to have to specify unique external FQDNs for each site i.e. webmailA.domain.com, webmailb.domain.com and then create a DNS webmail.company.com that you will point to either of the two sites. The users within that site will connect fine and the users in the other site will be redirected to the correct location... and rember certs, so make sure you have webmail.company.com on each internet facing CAS no matter in which site they are or you will have issues with site failures. You can also look at a geo-loadbalancer if you can define the areas by region. Hope this helps. Casper Pieterse, Principle Consultant - UC, Dimension Data South Africa, Microsoft Certified Master: Exchange 2007 / 2010

Thank you for the info. I'm not sure I understand everything completely though. We will have a geo-load balancer in place. IF I understand you correctly, you're saying that we should use a single FQDN to give to users, but that FQDN will just be a C Name record for webmailA and webmailB. Then we just configure the databases individually to use webmailA or webmailB and when a user logs in to webmail, they will be redirected to webmailA or webmailB according to their database settings. Is that correct?

I don't fully undertand geo-load balancers, but yes, that sounds correct.Casper Pieterse, Principle Consultant - UC, Dimension Data South Africa, Microsoft Certified Master: Exchange 2007 / 2010