You can track and guide at 2000mm, if you use an OAG and a sensitive guider. But it can be hard finding guide stars for some objects and it's not as tolerant of guide errors as a short FL and wider FoV...which will also be brighter per unit area. If you decide to run a guidescope with your C8, you're much better off lowering the FL/FR using a reducer-flattener. Bring that 2000mm down to something a bit more manageable, like 700-1500mm.
You can also use an OAG on a short FL scope as well. The good thing about an OAG is there's less weight on your mount (no guidescope etc) and no flex in the guiding (if setup properly).
Why the Atik is better for the C8 at 2000mm is because it's pixels size is larger than the other cameras I mentioned. It has 7.4micron pixels, the others range from 5.4 to 6microns. Larger pixels mean more sensitive to photons and usually larger full well capacities, which means they can gather more photons in and produce more charge in the electron wells of their p-n substrate without blowing out the image, especially on bright objects like some stars or planets. Plus you have to be careful of either under or over sampling the pixels. That's why you try and match your scope's aperture and FL to the camera. Go and have a look at this....
CCD University.
You've basically got it right, though.
Although, you can get chips with lots of large pixels e.g. KAF16803, KAI11000, KAI4000/4022, KAF6303 etc. They all have 7-9micron pixels. However, they're much larger than the little Sony ICX285AL in the Atik314L+ and cost considerably more...both as chips and as setup in the cameras. For example, the Atik314 is about $1650, the Atik11000 (using the KAI11000 chip) is about $6500 for the OSC version.