What this tool does
In a replicated system with N copies of the data, a write doesn't have to reach every replica to be considered successful, and a read doesn't have to check every replica either. Instead, each requires only a quorum — a write quorum of W replicas, and a read quorum of R. This tool checks whether a given N/W/R combination guarantees strong consistency, and how many replica failures it can tolerate.
The W + R > N rule
If W + R > N, every possible write quorum and every possible read quorum share at least one replica — so any read is guaranteed to overlap with the most recent acknowledged write and see it. If W + R ≤ N, it's possible to pick a read quorum that misses every replica holding the latest write, so a read can return stale data. That's not a bug — it's a deliberate trade for faster reads or writes, as explained in CAP Theorem Explained and Eventual Consistency Explained.
FAQ
What's the difference between fault tolerance and availability fault tolerance?
Write fault tolerance (N − W) is how many replicas can be down while a write still reaches enough acknowledgements. Read fault tolerance (N − R) is the same for reads. Availability fault tolerance is the smaller of the two — the most replicas that can be down while the system still serves both reads and writes.
Why does the majority preset use ⌊N/2⌋ + 1 for both W and R?
It's the smallest W and R that still satisfy W + R > N for both operations simultaneously, which balances read and write latency and fault tolerance evenly — the default most quorum-based systems (Dynamo-style stores, Raft, etc.) use.
What happens if W + R equals N exactly?
That's still eventual consistency, not strong — the rule is a strict inequality. With W + R = N, a read quorum and a write quorum can be chosen to be completely disjoint, so overlap isn't guaranteed.
Is my data uploaded anywhere?
No — everything happens entirely in your browser. Nothing is sent to a server.
Try it yourself
Need unique IDs across those replicas instead? Try the Snowflake ID Generator or the Distributed ID Explorer.