cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Joe Stump <...@joestump.net>
Subject Re: Partition data - advantage and disadvantage
Date Mon, 28 Dec 2009 15:07:18 GMT

On Dec 28, 2009, at 5:43 AM, JKnight JKnight wrote:

>    ~ org.apache.cassandra.dht.RandomPartitioner,

The advantage of the random partitioner is that it randomly distributes your keys across the
cluster. This (theoretically) avoids key clustering on nodes. The big disadvantage is that
you can't do key range operations.

>    ~ org.apache.cassandra.dht.OrderPreservingPartitioner, and
>    ~ org.apache.cassandra.dht.CollatingOrderPreservingPartitioner.

They're both ordering partitioners. The difference is that one is binary order and the other
is collating order. The advantage of these is that you can do key range operations. The disadvantage,
because of the ordering, you could have keys clustered on specific nodes.

That being said, we've found that with proper token choices you can avoid a lot of the key
clustering issues.

--Joe
Mime
View raw message