Scaling out in Cassandra

1.Scaling out VS scaling up

Scaling up

• scaling up refers to converting existing infrastructure to better or more robust hardware

• This could mean adding storage capacity, increasing memory, moving to newer machines with more cores

Scaling out

• scaling out simply means adding more machines that roughly match the specifications of the existing machines

• Since Cassandra scales linearly with its peer-to-peer architecture, scaling out is often more desirable

2.Adding nodes without vnodes

• If possible, run repairs to ensure all nodes contain the most recent data

• Make sure Cassandra is installed, but do not start the process. If you use a package manager, be aware that Cassandra will start automatically, so you will need to stop the process before proceeding

• Edit cassandra.yaml file as follows sudo vim /etc/cassandra/cassandra.yaml

• Set the seeds,initial value,rpc_address,endpoint and auto_bootstrap values

• Start the Cassandra daemon on the new node as follows sudo service cassandra start

• Check the status of the bootstrap process as follows sudo nodetool netstats

3.Adding nodes with vnodes

• Same as adding nodes without vnodes but,instead of initial_tokens, set the value of num_tokens to a greater value than exisiting nodes

• Update the appropriate properties files

• If you're using the GossipingPropertyFileSnitch, add the cassandra-rackdc.properties file on each new node. If you have chosen the PropertyFileSnitch, you will need to update cassandra-topology.properties on ALL nodes (a restart is not required on existing nodes)

• For eg

  Larger values represent proportionally larger nodes in your cluster, with 256 being the default. If all your nodes are the same size, this default should be sufficient

• Keep track of the bootstrapping process as follows sudo nodetool netstats