What is rendezvous hashing?

Definition

Rendezvous hashing computes a score for every key-and-node pair and selects the highest-scoring node. All participants with the same node list make the same assignment without storing a separate lookup table.

When a worker joins or leaves, only keys whose winning node changes need reassignment. This makes the method useful for distributing repositories across replaceable workers without permanent primaries.

ELI5

Rendezvous hashing gives every item a repeatable score for each available worker and assigns the item to the worker with the highest score. Everyone using the same worker list reaches the same answer without maintaining a separate assignment table.

For example, repositories can be distributed across several replaceable workers. When one worker disappears, only the repositories that chose it need to move, while most assignments remain stable.

How is rendezvous hashing different from consistent hashing?

Both minimize movement when nodes change, but rendezvous hashing ranks nodes per key instead of placing nodes and keys on a hash ring.

Why use rendezvous hashing for repository workers?

It provides deterministic assignment and limited reshuffling as workers scale or fail, without maintaining a fixed primary mapping.

  1. Why Git Hosting Is Hard beside a flat diagram of replicated repository servers and a latency bottleneck
Definition card for rendezvous hashing: What is rendezvous hashing?

Rendezvous hashing assigns each key to the available node with the highest deterministic score, minimizing reassignment when nodes change.