Figured I’d finally put some CompSci stuff on here since I’ve just been warbling on about business and random crap.
There’s a research project in Berkeley that tries to tie up the various cluster computing platforms, and give a common layer below systems such as Hadoop and Dryad and allow them share resources, interact with running processes etc.
The paper looks interesting, with a posted result of around 20% increase over running separate instances, though I’ve got to spend some time to dive into the details.
I am a little hesitant, since they haven’t yet released the implementation behind the research. But there’s always hope.
Also, it appears I’ve actually had two visitors! Welcome, please turn the lights out when you’re done.
