Quorum Consistency in Cassandra

In this article, we will discuss how quorum consistency is helpful in Cassandra and how we can calculate it, and also discuss how quorum consistency works. 

What is Quorum Consistency? 

Quorum consistency is consistency in Cassandra for high mechanism and to ensure that how many nodes will respond when we will define the read and write consistency in Cassandra. In Quorum consistency a majority of (n/2 +1) nodes of the replicas must respond. In Quorum, we check the majority of replicas (which simply means the number of replication factors). for example, if we have a replication factor of 3 in 2 data centers then how many their replicas will be there. so, there will be 6 and the majority is 4. (total_sum_replicas/2 + 1). Generally, we use local quorum in most cases in which we need 2 nodes from the LOCAL DC to succeed when if there are three replicas each in 3 data centers. Now, here we are just going to define the CQL query for the same. let’s have a look.

CREATE KEYSPACE cluster1
with replication =  {'class' : 'NetworkTopologyStrategy', 
                     'DC1': 3, 'DC2': 3, 'DC3': 3}
AND DURABLE_WRITES = false; 

To verify the results used the following CQL query given below.

SELECT * 
FROM system_schema.keyspaces; 

Output: How QUORUM is calculated?

This is how we calculate quorum which simply means how many nodes will acknowledge.

Quorum = (sum_of_replication_factors / 2) + 1 

Quorum is equal to the sum of replication factors division by 2 and added 1. It’s because to make it a whole number. 

Quorum consistency: The sum of all the replication_factor settings for each data center is the sum_of_replication_factors.

Total_sum_of_replication_factor = DC1_RF + DC2_RF + DC3_RF+ ... +DC_RF 

How does Quorum consistency work?

If there are three nodes that are in majority and they have to respond in Quorum consistency then in the below-given diagram acknowledgment is showing that there are three nodes responding and at the time writing data then committed nodes showing that we committed our data into three nodes.

In the above  “Commit data” diagram if we are inserting data into a table and RF =3 that means data will be replicated on three nodes that are available. The commit operation is complete only after successfully replicating in at least 3 nodes. In the  “Acknowledgement” diagram, it shows that if we are reading data and there are a total of 5 nodes while applying the quorum consistency we get (5/2 +1)=3 nodes. Hence we need acknowledgment from 3 nodes before responding to a read request.