Failures
Network Partitions: Designing for Broken Links
A practical explanation of network partitions and why services need explicit rules when nodes cannot talk.
Cluster State
Membership: Who Is in the Cluster
Why cluster membership is a first class concept in distributed systems.
Ordering
Lamport Time: Ordering Without a Shared Clock
A simple explanation of Lamport time and how logical clocks help reason about event order.
Partitioning
Key Partitioning: Put Data in the Right Place
How key partitioning distributes data and load across nodes, and why the key choice matters.
Cluster Management
Join Protocols: How Nodes Enter a Cluster
How a new node joins a distributed cluster without receiving traffic too early.
Retries
Idempotency: Make Retries Boring
A practical explanation of idempotency and how it makes retries safe in distributed workflows.
Failure Detection
Heartbeats: Knowing a Node Is Alive
How heartbeats help systems detect failures without confusing slow with dead.
Cluster State
Gossip Protocols: Small Messages, Shared State
A simple explanation of gossip protocols and why they are useful for spreading cluster state.
Concurrency
Fencing Tokens: Stop the Old Owner
How fencing tokens protect shared resources when an old leader wakes up after a pause.