Was just reading through the code to get an understanding of the memory impact 
for secondary indexes. The index CF is created with the same memtable settings 
as the parent CF (in CFMetaData.newIndexMetadata). 

Does this mean that when estimating JVM heap size each index should be 
considered as another CF? I'll update the wiki with the answer 
http://wiki.apache.org/cassandra/MemtableThresholds 

Cheers
Aaron
 

Reply via email to