You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@cassandra.apache.org by MK <st...@gmail.com> on 2010/08/27 05:39:55 UTC

How to ensure data placement on a specific node in a cluster

Hello,

I have started playing with a cassandra cluster and want to do the
following - When I insert a value into the cluster by sending the
"set" command to a node in the cluster, I want to ensure that atleast
one copy of the data will be stored on this node. (So that if this
node drops out of the cluster for network failure reasons, I can still
see the data that is present on this node)

What is the simplest way of doing that?

I think that trying to tweak the hashing will be too difficult but can
I somehow control the replication so that one replica will be on this
node? (I am using 0.6 version but have just stumbled on the "Local"
replica strategy in the head version - is that supposed to do what I
outlined above?)


Many thanks for any help!
-AB.

Re: How to ensure data placement on a specific node in a cluster

Posted by Jonathan Ellis <jb...@gmail.com>.
You can't scatter data semi-randomly by connection like that
permanently, but https://issues.apache.org/jira/browse/CASSANDRA-876
will let you store it temporarily (for the life of your connection).

On Thu, Aug 26, 2010 at 10:39 PM, MK <st...@gmail.com> wrote:
> Hello,
>
> I have started playing with a cassandra cluster and want to do the
> following - When I insert a value into the cluster by sending the
> "set" command to a node in the cluster, I want to ensure that atleast
> one copy of the data will be stored on this node. (So that if this
> node drops out of the cluster for network failure reasons, I can still
> see the data that is present on this node)
>
> What is the simplest way of doing that?
>
> I think that trying to tweak the hashing will be too difficult but can
> I somehow control the replication so that one replica will be on this
> node? (I am using 0.6 version but have just stumbled on the "Local"
> replica strategy in the head version - is that supposed to do what I
> outlined above?)
>
>
> Many thanks for any help!
> -AB.
>



-- 
Jonathan Ellis
Project Chair, Apache Cassandra
co-founder of Riptano, the source for professional Cassandra support
http://riptano.com