🗓️ 27082025 2343
📎

etcd

ABSTRACT

a distributed, consistent, and highly available key-value store

• Commonly used in distributed systems to store configuration, metadata, and state
• kubernetes uses etcd to store:
  • Cluster state
  • Node information
  • Secrets, ConfigMaps
  • Role bindings, CRDs, etc.

Key Features:

• Strong consistency ✅ (linearizable reads/writes)
• High availability ✅ (replicated across nodes)
• Watch support ✅ (clients can subscribe to updates)
• Based on raft consensus algorithm ✅ (ensures cluster-wide agreement)

etcd + Raft in Action

When you use etcd:

• Every write request:
  1. Goes to the leader.
  2. Leader replicates the write to followers via Raft.
  3. Waits for majority acknowledgment.
  4. Commits the change > responds to client.
• Every read can be:
  • Linearizable > from the leader for strong consistency
  • Stale > from a follower for better performance

Why This Matters

Use Case	etcd’s Role	Raft’s Role
Kubernetes	Stores cluster state	Ensures cluster-wide state consistency
Service Discovery	Stores endpoints	Guarantees updates are atomic
Feature Flags / Config	Stores dynamic configs	Keeps all replicas in sync
Leader Election	Stores leader info	Ensures only one true leader

---

References