Our MPLS backbone consists of Juniper, cisco, and soon, Brocade routers. All run IS-IS on the interior. Currently all is LDP, but we are considering RSVP-TE to do better engineering. (It's a fairly large and convoluted network.)
I could see using your rb750gs on the edge for our server farms. It would reduce the impact of single-points-of-failures (SPF). For example, instead of a large (but redundantly connected) Juniper box serving 48 servers. We would have 16 rb750gs boxes serving them. Each rb750gs would have a pair of redundant upstream MPLS ethernet links, and the other three ports serving 3 downstream servers (16x3=48). Now, if a PE device fails, it only kills three servers, not 48. Big win.
But, your box doesn't do IS-IS, which pretty much kills the plan. Is this something you are working on?
OSPF is great for smaller networks, but it's area/stub architecture does not work at large scale. We have learned that the hard way .
A more trivial question: are 19" rack rails available for the rb750gs?
John