Using Quorum Consensus to get consistency

In this technique, the system’s nodes work together to process and store data, and decisions are made based on the agreement of a majority of these nodes

Assumptions

$N$ - Number of replicas $W$ - Write quorum. For a write operation to be considered as successful, write operation must be acknowledged from $W$ replicas. $R$ - Read quorum. A read operation is considered successful if atleast $R$ replicas respond to the read request.

In the following figure, if $W=1$, it means that the coordinator must receive at least one acknowledgement for the write operation to be considered as successful.

The configuration of $W$, $R$ and $N$ determines the tradeoff between consistency and latency. For instance, if $W=1$ or $R=1$, the operation is performed quickly since the coordinator only needs to wait for a response from only one of the replicas, but the consistency of the system will take a hit.

Different quorum configurations

• If $W=1$ and $R=N$, the system is optimized for a fast write
• If $R=1$ and $W=N$, the system is optimized for a fast read
• If $W+R>N$, strong consistency is guaranteed because there must be at least one overlapping node that has the latest data to ensure consistency
• If $W+R\le N$, strong consistency is not guaranteed

Related Notes

• Meaning of consistency in distributed systems
• CAP Theorem
• Replicating data on the hash ring

Sources

• (Done) Design A Key-value Store - Highlights