I've been going through this a lot lately and thought I'd share just in case. I don't (yet, but I will) know all the details because the Infrastructure Team owns the resources and "the problem". They have a DR site setup and they use some sort of SAN based replication to keep them in sync. It apparently has some sort of a log (file) and replicates things asynchronusly through the log as a buffer. Everything works great and fast except if the log gets too far behind (typical because of nightly jobs) and the replication goes into what is known as the "synchronus" mode. When it does that, not only do things have to commit on the local server but they also have to commit on the remote server. The connections to the remote server will always be slower than the internal fiber connections we have to the SAN and form the bottle neck. Considering that the system is already way behind when it goes into the synchronuns mode, you can just imagine how slow that makes things.

I'm definitely not sure if this has anything to do with your problem at all or not but you can bet my current problem doesn't show up as a "DR Replication Problem" through the system DMVs. If you have a similar thing going on, the answer won't be obvious either. Check everything and expect the unexpected.