Observations on DAG Availability

http://blogs.technet.com/b/exchange/archive/2011/09/16/dag-beyond-the-a.aspx

Interesting reading in a recent blog post from the Exchange Team.

Some observations on the article:

• The multi-role scenario is touted as superior, however, it did not take into account redundancy of CAS servers and the redundancy of mailbox copies.
  • The failure of one server in a single role scenario with multiple CAS and MBX would result in better availability than the loss of a multi-role server.
• In the "Shared Storage Scenario" section, while nothing is actually incorrect, it should be noted that storage arrays have a probability of failure that is much lower than the DAS/SATA availability that was discussed in the article.  So, while the conclusion that "database copies on the same storage system decreases overall service availability" is not incorrect, it should be noted that many EMC arrays already provide 99.999% availability.  Without noting that the availability rate is much higher for a SAN, it leaves the reader with an incorrect impression of lower availability for shared storage than there is in actuality.

Any other observations out there?