incubator-cassandra-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Avinash Lakshman <>
Subject Re: BinaryMemtable
Date Fri, 03 Apr 2009 04:14:33 GMT
That is what we used to load large amounts of data into Cassandra using M/R.
So we loaded around 12TB of data from Hadoop into Cassandra before we
launched Inbox Search. This way we could do all the heavylifting in Hadoop
and load data at practically network bandwidth 100 MB/sec. Going the normal
route with the same load chewed up lot of CPU resources on the Cassandra
servers because of lot of serialization/deserialization.

On Thu, Apr 2, 2009 at 9:07 PM, Jun Rao <> wrote:

> Could anyone explain the purpose of BinaryMemtable? From the code, it seems
> that it's used to store some binary data per key. However, it's not clear
> who is using it and for what purpose. Thanks,
> Jun
> IBM Almaden Research Center
> K55/B1, 650 Harry Road, San Jose, CA  95120-6099

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message