TECH SOLUTIONS

Cassandra is well suited when it runs in distributed mode in a cluster. It follows peer to peer cluster mode instead of master/slave architecture. That makes Cassandra really good in terms of availablility and scalability.

Suppose , I have two machines with ip "192.168.1.11" and " 192.168.1.24".We can create a cluster of these machines as follows:

On both of the systems do the following changes in cassandra.yaml file:

1) listen_address-

for example,

On first machine, listen_address-192.168.1.11

On second machine, listen_address-192.168.1.24

2)rpc_address-

Similarly as listen_address

3) Add both ips as seed nodes on each of the machine.

For example,

seed - "192.16.1.11,192.168.1.24"

Make the above change in both of the systems.

4) Start the cassandra deamon on both of the machines

$CASSANDRA_HOME/bin/cassandra -f 

5) Run the nodetool on any of the system to check as follows:

$CASSANDRA_HOME/bin/nodetool -h 192.168.1.11 ring

Note: Ownership information does not include topology, please specify a keyspace.

Address DC Rack Status State Load Owns Token

125770647146966923544277099442418897726

192.168.1.24 datacenter1 rack1 Up Normal 11.32 KB 49.51% 39864057911651858052466199027576462686

192.168.1.11 datacenter1 rack1 Up Normal 13.52 KB 50.49% 125770647146966923544277099442418897726

$CASSANDRA_HOME/bin/nodetool -h 192.168.1.24 ring

Note: Ownership information does not include topology, please specify a keyspace.

Address DC Rack Status State Load Owns Token

125770647146966923544277099442418897726

192.168.1.24 datacenter1 rack1 Up Normal 11.32 KB 49.51% 39864057911651858052466199027576462686

192.168.1.11 datacenter1 rack1 Up Normal 13.52 KB 50.49% 125770647146966923544277099442418897726

That defines how easily we can scale cassandra cluster. We can add another node similarly.

Note: Time and date on all the systems should be in sync.

Click here for Other topics of BigData Technologies