procs -----------memory---------- ---swap-- -----io---- --system-- -----cpu------ r b swpd free buff cache si so bi bo in cs us sy id wa st 5 1 29576 88400 27656 13797784 1 0 46 134 0 1 8 1 91 1 0
performance is definitely better now that caching is working, thx! however, the pending tasks in compaction pool are still not decreasing and i've reduced the load a bunch on the server. i'll be doing another test tonight with some modifications to my test client to simulate real world a bit better. On Fri, 2009-12-04 at 16:47 -0600, Jonathan Ellis wrote: > how much does vmstat say is being used for cache? > > if it's not most of the other 14, then read I/O doesn't sound like the culprit > > On Fri, Dec 4, 2009 at 4:17 PM, B. Todd Burruss <bburr...@real.com> wrote: > > thx for the tip. i've allocated -Xmx2G for cassandra out of 16G. > > > > > > On Fri, 2009-12-04 at 14:22 -0600, Brandon Williams wrote: > >> On Fri, Dec 4, 2009 at 1:40 PM, B. Todd Burruss <bburr...@real.com> > >> wrote: > >> i've actually allocatted plenty of RAM, but it isn't fully > >> used. the > >> read performance dwindles well before reaching a RAM limit. > >> > >> > >> You don't want to over-allocate memory to Cassandra, you want to leave > >> as much free as you can so the OS can cache the reads for you. > >> > >> > >> -Brandon > >> > > > > > >