Blue-Green vs Canary Deployment: Shipping Without Downtime
Blue green vs canary deployment decided: zero downtime deployment mechanics, the canary release strategy and its observation ramp,…
Tag
system-design
58 articles
Retry with Backoff and Jitter: Retrying Without Triggering Retry Storms
Retry with exponential backoff explained: backoff and jitter pacing, the retry storm math, retry budgets, what is safe…
Disaster Recovery: RTO, RPO, and Getting Back Online
Disaster recovery explained: RTO and RPO as the two numbers that size the plan, the disaster recovery strategies…
High Availability: Designing Away Single Points of Failure
High availability explained: the single point of failure hunt, n+1 redundancy sizing, failover design that does not become…
Backpressure: Managing Load You Cannot Process
Backpressure explained: the flow-control signal that travels from a slow consumer back to its producers; bounded queues where…
Load Shedding: Dropping Traffic to Stay Alive
Load shedding explained: overload survival by dropping work deliberately: graceful degradation ladders chosen in advance, server overload protection…
Idempotency: Designing Operations That Survive Retries
Idempotency explained: operations that survive retries and duplicates, the idempotency key from client to dedup table, idempotent API…
The Saga Pattern: Distributed Transactions Without Two-Phase Commit
The saga pattern explained: long-lived business transactions as local steps with compensating transactions, orchestration vs choreography, the transactional…
Distributed Locks: Mutual Exclusion Across Machines
Distributed locks explained: leases and TTLs as mutual exclusion across machines; the Redis distributed lock and its failover…
Fault-Tolerant Systems: Redundancy, Failover, and Failure Detection
Fault tolerant system design: failure models from crash to byzantine, redundancy and the N+1 arithmetic, the failover playbook,…
Distributed Transactions: Two-Phase Commit and Its Alternatives
Distributed transactions explained: two-phase commit step by step, the 2pc coordinator failure that leaves participants in doubt, consensus-backed…
Leader Election in Distributed Systems: Methods and Split-Brain Risk
Leader election in distributed systems explained: leader election algorithms from the bully algorithm to quorum voting, zookeeper leader…