liveBook · Manning

Linearizability

As explained in ScyllaDB in Action

Linearizability is a consistency property in which each operation appears to occur atomically in an order that respects real-time ordering.

Linearizability makes each read appear atomic and orders reads according to when they occur in real life. In ScyllaDB, SERIAL consistency uses Paxos to read the latest value, but its stronger guarantee costs more. section 8.2.2

The consistency choices create practical trade-offs:

  • SERIAL consistency uses a majority of replicas and Paxos to enforce read ordering. section 8.2.2
  • QUORUM consistency for both writes and reads can provide the latest value without an LWT-based read. section 8.2.2

Choosing a read guarantee

Use SERIAL when a read must reflect the latest value and its Paxos cost is acceptable. Otherwise, pair QUORUM writes with QUORUM reads to avoid stale results through replica majority.

How ScyllaDB in Action covers it

How deep it goes
Defines linearizability through ordered latest-value reads, then defers detailed Paxos costs and operation.
Examples
Uses SERIAL read ordering, stale-read figures, and QUORUM reads and writes as concrete illustrations.
What you can do after
Apply ScyllaDB SERIAL or QUORUM consistency according to latest-value guarantees and Paxos cost.
How it connects
Links linearizability to consistency levels, SERIAL consistency, Paxos, quorum consistency, and stale reads.

Also explained in

Compare how 4 books explain linearizability →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of ScyllaDB in Action. The section links point to the parts of the book it draws on.