Basically we run 2 instances on one HP 980 for Dev and QA , our prod is a 980 by its self - when we went for the licensing we took the unlimited version , however we made a decision to only keep the last two years data in the system , and we use our old BW (on SQL) as a store - now that means we run parallel process chains, however 0 performance issues in ECC. We had to get the system data (at a point in time) the same - then we left the old BW PC running once a day , in hana we run "near time" every 5 mins - I freaking love it. with our SOX audit they asked us top dump BKPF and BSEG for the current fiscal into some txt files (for us that was 38 million rows) we already had the join created in BW - extracted 12 files in about 30 mins , including creating teh selection criteria every pull . so yes it works , apparently we will be able to make it even faster once we optimize the structures for hana.
If you want to know more detail PM me and we can arrange a call