http://wiki.apache.org/cassandra/CassandraLimitations
"Cassandra's compaction code currently deserializes an entire row (per columnfamily) at a time. So all the data from a given columnfamily/key pair must fit in memory. Fixing this is relatively easy since columns are stored in-order on disk so there is really no reason you have to deserialize row-at-a-time except that that is easier with the current encapsulation of functionality. This will be fixed in https://issues.apache.org/jira/browse/CASSANDRA-16"; so yes, based on the amount of data you have in there and your memory allocated to cassandra there is a limit currently...hopefully gone in 0.7 I will note that another issue I run across with this is that get_count will not return within the rpc timeout after you get a certain amount of columns in there. cheers, jesse -- jesse mcconnell jesse.mcconn...@gmail.com On Wed, Mar 24, 2010 at 17:05, Davis, Jeremy <jda...@gridpoint.com> wrote: > > > Hello, > > Is there a practical limit on the number of columns I put on a key? > > Obviously if I tried to grab the entire row at once I would have a problem. > > However, if I had an open ended row, with column names of “1” to “999999999” > etc. And I only accessed ranges, would there be a practical limit I would be > running up against? Performance degradation over time, etc? > > > > I’m running a little micro benchmark to see how well this works, but just > curious if anyone has any insights. > > > > Many Thanks, > > -JD
