On Fri, 30 Apr 2010 11:30:28 -0500 Paul Krizak <paul.kri...@amd.com> wrote:
PK> The main thing I'm asking is if anybody has had any experience scaling PK> up a cfservd to this much work, and whether or not it fell over. ... PK> I'm talking about 5,000 systems each updating about 50MB of policy data PK> over a 1-2 hour span. I've been interested in Cassandra, which is an efficient distributed database. It may be of interest to you as well. It has C bindings and could serve as the backend of a distributed cfservd. It scales nearly linearly with the number of servers so accomodating your load is a simple matter of adding more servers. The main philosophical point is that Cassandra (like Amazon's Dynamo) is eventually consistent, unlike the standard cfservd backend and traditional RDBMSs. I'd be interested in any thoughts from Mark Burgess and others on how eventually consistent policy distribution methods can fit in with the cfengine convergence model (if any existing literature already addresses this, I'd love to find it). Ted _______________________________________________ Help-cfengine mailing list Help-cfengine@cfengine.org https://cfengine.org/mailman/listinfo/help-cfengine
