As the FB guys seem to be a bit busy, I'll take a stab at answering this. Firstly cassandra uses a gossip protocol for dynamic addition (and removal) of servers from the pool. when you bring up a new node you simply configure a set of seeds for it to initially communicate to and your away.
It is using eventual consistency (see Warner Vogel's explanation http://www.allthingsdistributed.com/2007/12/eventually_consistent.html for an intro to this) to do replication across the pool. I believe this is different than hbase/hypertable. It is more suited to OLTP than to batch environments than hbase/hypertable. HTH (and is that it's correct ;-) Ian On Fri, Dec 5, 2008 at 3:08 AM, Otis Gospodnetic <[EMAIL PROTECTED] > wrote: > Hello, > > The "distributed storage system for managing structured/unstructured data > while providing reliability at a massive scale." part sounds kind of like > HDFS. Would it be possible to describe how Cassandra is different from > HDFS? Perhaps the best place to do it is under the "Relationships with > Other Apache Products" section. > > Thanks, > Otis > -- > Sematext -- http://sematext.com/ -- Lucene - Solr - Nutch > > > > ----- Original Message ---- > > From: Avinash Lakshman <[EMAIL PROTECTED]> > > To: "general@incubator.apache.org" <general@incubator.apache.org> > > Cc: Prashant Malik <[EMAIL PROTECTED]>; Kannan Muthukkaruppan < > [EMAIL PROTECTED]> > > Sent: Monday, December 1, 2008 7:51:28 PM > > Subject: Cassandra Incubator Proposal > > > > Hi Folks > > > > Please consider our proposal to move the Cassandra project into the > Incubation > > process - http://wiki.apache.org/incubator/Cassandra. Please advice, as > to what > > else is required for us complete this process. > > > > Cheers > > Avinash > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > >
