On Wed, Feb 16, 2011 at 7:12 AM, Victor Kabdebon <victor.kabde...@gmail.com> wrote: > Someone please correct me if I am wrong, but I think the overhead you can > expect is something like : >
MemTableThroughtPutInMB * <JavaOverheadFudgeFactor> * <maximum likely number of such memtables which might exist at once, due to flushing logic> JavaOverHeadFudgeFactor is "at least 2". The maximum likely number of such memtables is usually roughly "3" when considered across an assortment of columnfamilies with different write patterns. > but I don't know when BinaryMemTableThroughputInMb come into account.. BinaryMemTable options are only considered when using the Binary Memtable interface. If you don't know what that is, you're not using it. =Rob
