On Sat, Feb 12, 2011 at 11:17 PM, E S <tr1skl...@yahoo.com> wrote: > While experimenting with this, I found a bug where you can't have memtable > throughput configured past 2 gigs without an integer overflow screwing up the > flushes. That makes me feel like I'm in uncharted territory :).
I am sure the project would appreciate a JIRA ticket detailing how to reproduce this behavior, which sounds like a bug. https://issues.apache.org/jira/browse/CASSANDRA Regarding very large memtables, it is important to recognize that throughput refers only to the size of the COLUMN VALUES, and not, for example, their names. In cases where, for example, an empty column value is being stored with a UUID column name, your memtable can be much larger than the value in the cf definition. I have seen 1GB-sized memtables flush to disk as 1.7GB, and they are likely significantly larger in memory due to Java object overhead. =Rob
