You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@cassandra.apache.org by nicolas lattuada <ni...@hotmail.fr> on 2010/12/27 18:08:14 UTC

How to setup a big shared files system

Hi

I am working for an afiliate marketing program, we are logging a lot of data, and thinking about using Cassandra.

But i wonder which is the way to setup properly a cassandra cluster.

For example let s say i would like to store about 100 000 Giga bytes of data, and will need to access them quite fast sometimes.


Of course the data cannot stand on a single server, so what would you recomend for these needs?

Is it possible to have multiple cassandras instances accessing the same index over a shared file cluster, or is it better to do partitionning?

Is  hardware for such a setup expensive?

thank you.
 		 	   		  

Re: How to setup a big shared files system

Posted by ArtemGr <ar...@gmail.com>.
nicolas lattuada <nicolaslattuada <at> hotmail.fr> writes:
> Of course the data cannot stand on a single server,
> so what would you recomend for these needs?

Run multiple Cassandra instances, one instance per server.

> Is it possible to have multiple cassandras instances
> accessing the same index over a shared file cluster,
> or is it better to do partitionning?

AFAIK, it should be possible
to run multiple Cassandra instances over a POSIX-compatible cluster filesystem,
but this would not be efficient.

> Is  hardware for such a setup expensive?

Take, for example, server4you servers
(http://www.server4you.com/root-server/ecoserver.php)
Your 100 000 Giga bytes / 320 gb * 2 (replication factor) * 1.5 (overhead)
= a thouthand servers? 39000$ / month?