In regard to the column lifetimes:
Not every sub-2 is equal. Stick to the companies that have been manufacturing them for a while, like Agilent and Waters. I've used two columns from two other column manufacturers (besides Waters and Agilent) - one a true sub-2, the other a 3-micron running at high pressure (ostensibly within the acceptable pressure range for the column), and had them die within about 100-125 injections during method development. Meanwhile, I have an Agilent ZORBAX RRHD Extend C18 3.0x100mm 1.8-micron that I've been running for months at 650-750 bar with real environmental samples and no guard column, just a 0.2-micron in-line Phenomenex KrudKatcher filter, and I've had no problems in terms of column lifetime. There are probably 1000-1500 injections on this column by now, which to me is well within the range of acceptable lifetime for any column, especially without a guard column.
A caveat about new instruments in general, and this is my experience only, so I can't say that you'll encounter the same problems, or even need to consider the same issues, but I want to say my piece anyway...
1. Make sure there are an adequate number of service personnel in your area before you buy whatever you end up buying.
2. Get a service contract. It might take you a while to learn the new instrument, and the difference between breaking it due to your own ignorance while within the coverage of a service contract and outside the service contract can be substantial. Also, some manufacturers are so new to UHPLC/Fast LC/whatever the new euphemism is, that their hardware is not fully tested in the ways that a real chromatographer will test it (read: break it). Trust me - I have an Agilent engineer coming out Monday to replace the pump drives on one of my Agilent 1290 binary pumps (not my fault, I promise - but I'm glad it's under contract).
3. Make sure your new system is compatible with your old data quantitation/acquisition software. If not, make sure that any new software you might need fits into your current workflow scheme - you don't want to get a fancy new instrument, only to realize the data it generates is incompatible with either your LIMS system or your quantitation (quantification?

) program.
[hijack][/end hijack]