Designing for CAS and Site Resilience... [some more thoughts]

Just carrying on from my previous blog....  Exchange 2007 has been designed to integrate even more closely with AD, a central part of which is AD sites.  Therefore it is most logical to assume that your preferred design is that which also makes sense for other applications and services that use AD sites.  E2K7 has been designed so that most companies should not have to alter their current AD Site topology in order to accommodate it.  So unless you use dedicated AD sites for Exchange which is not the preferred option then it makes sense that Exchange is designed to fit as neatly as possible within the current site infrastructure.  This makes design 2 (or possibly design 1) in my previous blog the most obvious choice and seems to be the design of choice of the product group; given this particular set of requirements.

So   ...to oversimplify the issue for the purposes of this blog, administrators must choose to proxy all CAS traffic from outside, and have a cas proxy tier and a single namespace; or avoid CAS proxying by removing the CAS tier but live with multiple regional namespaces.

I think that a large company which does actually operate out of regional centres, which their user community are logically tied to, might see the use of multiple namespaces as a good viable option.  ...but I think companies in the UK for example where data centres might be relatively close together (and who's location might be unknown to the user community even) might consider regional namespaces unworkable to achieve what they want because they won't make any sense to the user community.   They might then feel forced down the CAS proxy tier route as in design 1  ...but then to proxy everything sounds expensive in terms of performance and extra hardware, and it adds a layer of complexity which may become expensive operationally...?