What is quorum and why is it critical in distributed systems?

What is quorum and why is it critical in distributed systems?

• A. The minimum number of agreeing nodes needed to proceed, preventing split-brain and ensuring data consistency.
• B. The total number of nodes in the cluster.
• C. The speed at which nodes respond to requests.
• D. The maximum delay allowed for a transaction.

Answer: (A)

Quorum is the minimum number of nodes that must agree before the system can move forward with a decision. This threshold is chosen so that any two agreeing groups overlap by at least one node, which keeps decisions consistent even if parts of the system are partitions away from each other. That overlap is what prevents split-brain and protects data integrity: a write that is acknowledged by a quorum cannot later be contradicted by another partition making a competing decision, because the two majorities share at least one node and a single agreed state. In practice, a cluster might use a majority as its quorum (for example, in a five-node cluster the quorum is three), so as long as at least three nodes agree, the system can commit changes and continue operating; if too many nodes fail or become partitioned away, progress halts to avoid conflicting states. This concept is not the same as the total number of nodes in the cluster, which is just the size of the system. It is not about how fast nodes respond or the maximum delay for a transaction, which relate to performance and latency rather than the safety threshold that ensures consistent progress.

Quorum is the minimum number of nodes that must agree before the system can move forward with a decision. This threshold is chosen so that any two agreeing groups overlap by at least one node, which keeps decisions consistent even if parts of the system are partitions away from each other. That overlap is what prevents split-brain and protects data integrity: a write that is acknowledged by a quorum cannot later be contradicted by another partition making a competing decision, because the two majorities share at least one node and a single agreed state. In practice, a cluster might use a majority as its quorum (for example, in a five-node cluster the quorum is three), so as long as at least three nodes agree, the system can commit changes and continue operating; if too many nodes fail or become partitioned away, progress halts to avoid conflicting states.

This concept is not the same as the total number of nodes in the cluster, which is just the size of the system. It is not about how fast nodes respond or the maximum delay for a transaction, which relate to performance and latency rather than the safety threshold that ensures consistent progress.