Thanks for the links and additional explanation Chris. I will go through them.
One more question. My hosting provider sets a hard limit on the harddisk size at the time of order. I go with, say, 20TB harddisk per server and order 5 of them, I can get Riak running but I will still be limited to 20TB. If I need to increase my harddisk space, I would have to upgrade my hardware that allows more disk bays. (limitation of my provider). So if I needed to increase the capacity to, say, 60TB, will it work if I add 5 more 60TB servers to the existing pool and decommission the old ones after the data are replicated? How long does it take and how do I know the replication is complete in the newly joined nodes? On Fri, Apr 3, 2015 at 11:59 PM, Christopher Meiklejohn < [email protected]> wrote: > > > On Apr 3, 2015, at 11:24 AM, Shankar Dhanasekaran <[email protected]> > wrote: > > > > Thanks Chris for your answer. I still don't comprehend what is meant by > > this is the minimum amount of nodes needed to ensure all copies of each > object are stored on different physical nodes. [2] > > > > Why can't a 3 node cluster store the data in each of them? I am reading > the little riak book as well but still I don't understand what is the magic > value of 5? > > Five is the minimum number of nodes required to guarantee each copy of > each object is not stored on the same physical machine. With three this > can not be guaranteed, so you might have two copies (out of three) for a > given object on the same machine. > > It’s an optimization problem computed during node membership referred to > as “claim.” > > [1] > http://lists.basho.com/pipermail/riak-users_lists.basho.com/2011-February/003319.html > [2] > http://basho.com/why-your-riak-cluster-should-have-at-least-five-nodes/ > > Christopher Meiklejohn > Senior Software Engineer > Basho Technologies, Inc. > [email protected] > >
_______________________________________________ riak-users mailing list [email protected] http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com
