The max size would probably be best determined by looking at the size of your MemTable
<!-- ~ Flush memtable after this much data has been inserted, including ~ overwritten data. There is one memtable per column family, and ~ this threshold is based solely on the amount of data stored, not ~ actual heap memory usage (there is some overhead in indexing the ~ columns). --> <MemtableThroughputInMB>64</MemtableThroughputInMB> Read repair is on a per column basis, every column gets a timestamp, and the overhead of a name. So, balance those 3 out and you have a pretty good idea of what to do. From: Dop Sun [mailto:su...@dopsun.com] Sent: Thursday, April 29, 2010 7:38 AM To: user@cassandra.apache.org Subject: RE: What's the best maximum size for a single column? Is there any practical number can refer to? Like what's the size (big one) used in single columns in your application? From: uncle mantis [mailto:uncleman...@gmail.com] Sent: Thursday, April 29, 2010 1:57 AM To: user@cassandra.apache.org Subject: Re: What's the best maximum size for a single column? There is no column size limitation. As to performance due to the size of a column and with the speeds that Cassandra are running at, I don't belive it would make a bit of a difference if it was 1 byte or a million bytes. Can anyone here prove me right or wrong? Regards, Michael On Wed, Apr 28, 2010 at 7:37 AM, Dop Sun <su...@dopsun.com<mailto:su...@dopsun.com>> wrote: Hi, Yesterday, I saw a lot of discussion about how to store a file (big one). It looks like the suggestion is store in multiple rows (even not multiple column in a single row). My question is: Is there any best maximum column size which can help to make the decision on the segment size? Is this related with the memory size or other stuff? Thanks, Regards, Dop
