The answer will likely depend on how much of the work you can do yourself (DNA extraction, Library Preparation, Data Analysis) and how much you need to outsource. It would be interesting to contact some of these platforms and get quotes:
The most important is to know if the bacteria are cultivable. If it is, MiSeq of HiSeq2000/2500 should be enough. Yo can start with pair-end library (and decide to do a mate-paired library denpending in your repetitive regions) or just perform a pair-end and a mate-paired library at the same time.
We worked wit Macrogen (http://www.macrogen.com), that has very good prices and is a serious company.