On Mon, Jan 31, 2011 at 2:49 PM, Chunqiang Tang <ct...@us.ibm.com> wrote: > The paper [2] below from VMware is informative but cannot be adopted by us > directly, as the problem domain is different. I previously had a paper on > general congestion control, > https://sites.google.com/site/tangchq/papers/NCI-USENIX09.pdf?attredirects=0 > , and attended a conference together with an author of paper [2] , and had > some discussions. It is an interesting paper. > > [1] http://suif.stanford.edu/papers/nsdi05.pdf . > [2] http://www.usenix.org/events/fast09/tech/full_papers/gulati/gulati.pdf
Thanks for the links, I will take a look. I can see both latency and throughput being problematic. With throughput the problem is that the I/O pattern may be sporadic or bursty. It's hard for a heuristic to follow large changes in the guest's I/O pattern. Stefan