cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From TuxRacer69 <tuxrace...@gmail.com>
Subject disk usage management and scaling
Date Sun, 08 Nov 2009 20:04:06 GMT
Hello Everybody,

I am trying to understand how disks are filled in cassandra.

If we run cassandra in a cluster of N (commodity) servers and each 
server has the same DataFileDirectories configuration

  <DataFileDirectories>
      <DataFileDirectory>/var/lib/cassandra/data</DataFileDirectory>
  </DataFileDirectories>

and with the same disk size, what should I do to never get a disk full 
on those servers?
Does cassandra scaling only me to act like follows:

I just watch the percentage of use of the partition containing  the 
/var/lib/cassandra/data on each server and if one of the servers returns 
a usage greater than a threshold (say 95%), then I just have to add an 
extra N+1 node to my cluster?
Will the disk usage eventually stablize a the average disk usage 'U' it 
was before the node addition to the lower value 'U * (N/N+1)'?
Is that that easy? ;)

Thanks
Alex


Mime
View raw message