I'm posting to this list hoping for feedback on a couple of disaster recovery
options we're exploring.

We have a three-tier (wfe, app, database) farm environment of SP 2010. In our
production environment, we have an offsite data center housing DR wfe, app, and
database servers. We are trying to determine the best way to configure that
environment. Our options are:

1. Synchronously mirror the database servers. Database failover happens
automatically in SQL and SharePoint.

2. Asynchronously mirror the database servers. Failover occurs with
intervention at the web front ends. Only content and certain service
applications are mirrored. Two farm environments need to be maintained.

This seems like a no brainer. The first option is easier to establish, easier
to maintain, and generally a tighter solution. However, almost all
documentation says that the link required between data centers has to be <1ms.
The SharePoint 2010 Disaster Recovery Guide lists this as a "recommendation".
My question is, how hard and fast a rule is this? What are the problems that
could manifest by having a slower link? Are we talking about mere page load
issues by waiting for SQL changes to commit, or possible commit queue problems
resulting in database errors?

Our networking group reports the link as between 2 and 3 ms. I've done ICMP
testing for extended periods (ping) and while I've seen outliers at 2 and 3 and
as high as 9, the extended average has been <1.