Engineering guidelines for replication, consistency, and consensus in distributed systems — with a complete design review checklist covering failure design, consistency model selection, replication configuration, consensus placement, performance, and observability.
Performance trade-offs in distributed replication and consensus — write latency, tail latency, write vs read scalability, consensus throughput limits, batching and pipelining, geographic distribution costs, and backpressure design.
How quorums work in distributed systems — the W + R > N rule explained with worked examples, strict vs sloppy quorums, read repair and practical configuration guidance for Cassandra, DynamoDB and Riak.
Leader-based, multi-leader, and leaderless replication explained — with synchronous vs asynchronous replication, replication lag, quorum configuration, and real production examples from PostgreSQL, Kafka, Cassandra, and Google Spanner.
CAP is not a design choice you make once — it is a constraint that surfaces when the network fails. This post explains CAP correctly, debunks common myths, introduces PACELC, and gives engineers a practical framework for applying CAP thinking per operation.
Consistency models are contracts between a system and its clients. This post explains the full spectrum from linearisability to eventual consistency — what each guarantees, what it costs, and when to use it.