incubator-cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Oleg Tsvinev <oleg.tsvi...@gmail.com>
Subject Cassandra 0.7.* replication question
Date Thu, 17 Mar 2011 17:54:29 GMT
I wonder what it the right way to configure replication in Cassandra cluster.

I need to have 3 copies of my data in a cluster consisting of 6 nodes.
3 of these nodes are in one datacenter - let's call it DC1 - and 3 in
another, DC2. There is a significant latency between these datacenters
and originally my application is going to be configured to only talk
to 3 local cassandra nodes, without auto-discovery.

It looks like I can use
org.apache.cassandra.locator.NetworkTopologyStrategy for a
placement_strategy in keyspace definition and provide values for
strategy_options from
$CASSANDRA_HOME/conf/cassandra-topology.properties. It does not seem
to be well-documented and I would like additional examples to
configure this option.

However, org.apache.cassandra.locator.NetworkTopologyStrategy is not
flexible enough because it requires changes in
$CASSANDRA_HOME/conf/cassandra-topology.properties every time the
cluster configuration changes, and that requires changes in keyspace
definition as well. Thus does not seem to scale well.

Is there elaborated examples for
org.apache.cassandra.locator.NetworkTopologyStrategy and keyspace
configuration?
Do I have any other options for my configuration?

Thank you,
  Oleg

Mime
View raw message