I have 1 RB2011UAS-IN and 1 OmniTIK-U-5HnD. For some reason, these do not seem to play nice together. When connecting the OmniTik to the RB2011 I'm only getting a throughput of about 3-4MB/s, connecting the OmniTik to and Airport or the RB2011 to the client I'm getting >10MB/s. Only when both RB2011 and Omnitik have traffic flowing through them is when it get's buggered up.

Here are some scenarios (note; ether2-5 all have ether1 as master port, i.e. one big switch group):
=== is cable
~~~ is wireless

via rb2011 but without omnitik:
server === rb2011 === client: 10MB/s
server === rb2011 === airport ~~~ client: 10MB/s

via rb2011 and via omnitik:
server === rb2011 === airport === omnitik ~~~ client: 3MB/s
server === rb2011 === omnitik ~~~ client: 3MB/s

via omnitik but without rb2011:
server === airport === client: 10MB/s
server === airport === omnitik ~~~ client: 10MB/s

So only when traffic flows through both omnitik and rb2011, throughput drops significantly. taking either out of the equation solves it. This also applies when substituting the omnitik for a groove-2Hn or groove-5Hn. (All devices run 6.0rc11).
Where do I begin to look?

Bridging or routing? Try alternating configs to see if there is a difference.