As I understand, overhead of Erlangs' fully connected node network is in TCP pings or heartbeats that nodes send to each other. Each node pings every other node that it is connected to. So that ping overhead grows quadratically with node count.
Frequency of those heartbeats can be set with -kernel net_ticktime parameter. I wonder did you tweak that parameter in some way with Riak? Perhaps some other tweaks? Regards, Sharas On Sat, Jan 28, 2012 at 11:31 PM, Anthony Molinaro < antho...@alumni.caltech.edu> wrote: > We've got a riak_core based application in production which currently has > 83 and which at one point had 85 nodes so the practical limit is definitely > higher than 20. It would be nice to know of a real limit though... > > -Anthony > > On Jan 28, 2012, at 7:31 AM, David Smith <diz...@basho.com> wrote: > > > On Sat, Jan 28, 2012 at 8:18 AM, Sharas gpavarde <apkro...@gmail.com> > wrote: > >> Hi > >> > >> I wonder does Riak cluster maintain fully connected Erlang network > between > >> machines? > >> As far as I know Erlang node network has a practical limit of around 20 > >> machines > >> Does this limit influence Riak? > > > > Hi Sharas, > > > > Erlang/Riak does indeed maintain a fully-connected network between > > nodes. The limit is much higher than 20 nodes, however -- anecdotal > > evidence suggests that 200+ nodes is not out of the question. > > > > D. > > > > -- > > Dave Smith > > Director, Engineering > > Basho Technologies, Inc. > > diz...@basho.com > > > > _______________________________________________ > > riak-users mailing list > > riak-users@lists.basho.com > > http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com >
_______________________________________________ riak-users mailing list riak-users@lists.basho.com http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com
