Dear Wiki user,

You have subscribed to a wiki page or wiki category on "Cassandra Wiki" for 
change notification.

The "CloudConfig" page has been changed by FlipKromer.
The comment on this change is: Described what I've found on EC2-optimized 
configuration, and added some questions for a later expert.
http://wiki.apache.org/cassandra/CloudConfig

--------------------------------------------------

New page:
= Setting up Cassandra in the Cloud =

''If you have done work to optimize your Cassandra install in the cloud, please 
take a moment to contribute some of that knowledge to this page'' 


== Amazon Web Services (AWS/EC2) ==

 * There is an ec2snitch to make Cassandra rack-aware in the ec2 cloud.
 * [[http://github.com/b/cookbooks/tree/master/cassandra|Chef install for 
Cassandra]], including ec2snitch setup.

=== Optimizing Volume Performance for a Transient Cluster ===

Depending on node size, and on how many EBS volumes are attached, most EC2 
nodes will have many independent attached volumes. 

 * How should the Cassandra config be modified to take advantage of multiple 
attached volumes?
 * What are the tradeoffs for EBS vs local drives as backing store for a 
persistent cluster?

Cloud clusters are often transient: when all jobs are finished, the nodes are 
terminated. Any data held on EBS volumes will remain until next restart, and 
any data on the local scratch disks disappears forever. When the cluster 
restarts, each node will have a new ip addresses and identity.

 * For a non-persistent cluster, can Cassandra take advantage of the scratch 
disks (assume they are fast but could disappear at once across the whole 
cluster at any time)

Reply via email to