incubator-cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Anthony Ikeda <anthony.ikeda....@gmail.com>
Subject Trying to understand QUORUM and Strategies
Date Wed, 31 Aug 2011 19:07:24 GMT
Okay, we are looking at setting up a production environment which means
getting our quorum settings and strategies correct. However, we need to
really understand the approach taken to get this right. So far we have been
working with co-located nodes and our prod environment is going to be
distributed across 2 DCs.

Our setup:
Dev = 2 nodes, SimpleStrategy, RF=1
QA = 3 nodes, SimpleStrategy, RF=2
Prod = 2 DC's (3 nodes each), NetworkTopologyStrategy, RF=3

For the Dev and QA environment we have been using ConsistencyPolicy of ONE
for reads and the default for writes

For Prod I'm guessing we will be using a ConsistencyPolicy of LOCAL_QUORUM
for both reads and writes.

Correct me if I'm wrong but as I understand it:

The Dev and QA environment can only use ConsistencyLevel of ONE or QUORUM.
LOCAL_QUORUM cannot be used - well we've seen an exception stating
LOCAL_QUORUM cannot be used with SimpleStrategy.

The Prod environment should be able to use all ConsistencyPolicy's, however,
we would have:
Reads: LOCAL_QUORUM
Writes: LOCAL_QUORUM


   - Is having LOCAL_QUORUM for reads going to affect us if we need to rely
   on the other DC as failover?
      - i.e. if we lose 2 nodes in DC1, will we still be able to read from
      DC2?
   - Would we need to ensure that any snitches are configured for the prod
   environment?
   - Is there anything else that I may be overlooking aside from these
   configuration values?


Anthony

Mime
View raw message