Taming Tail Latency with Hedged Requests
Learn how to cut p99 latency in APIs and AI agents with hedged requests: tuning the hedge delay, hedging only idempotent work, and capping the extra load.
Topic
reliability
18 Devspedia articles tagged with reliability.
Resource Budgets for Tool-Using AI Agents
Learn how to stop runaway AI agents with token budgets, cost ceilings, step limits, wall-clock deadlines, loop detection, and graceful degradation.
Compensating Actions for Tool-Using AI Agents
Learn how to make AI agent side effects safer with compensating actions, recovery policies, idempotent undo steps, and operator-ready audit trails.
Agent Command Ledgers for Reliable AI Workflows
Learn how to make AI agent side effects recoverable with command ledgers, fenced execution, reconciliation jobs, and replay-safe workflows.
Sandboxing Tool-Using AI Agents
Learn how to run tool-using AI agents behind capability manifests, policy gates, sandboxes, audit logs, and recovery controls.
Designing Durable Agentic Workflows
Learn how to design agentic workflows that survive retries, crashes, tool failures, human approvals, and partial progress without losing intent.
Designing Circuit Breakers for Distributed Services
Learn how to stop cascading failures with circuit breakers that open on real dependency pain, probe recovery safely, and expose clear fallbacks.
Designing Bulkheads for Resilient Services
Learn how to isolate service capacity with bulkheads so one slow dependency, tenant, queue, or feature cannot exhaust the whole system.
Designing Load Shedding and Backpressure for APIs
Learn how to protect APIs during overload with admission control, bounded queues, backpressure signals, and clear degradation rules.
Designing Retry Strategies with Backoff and Jitter
Learn how to retry transient failures without amplifying outages by combining timeouts, backoff, jitter, budgets, and observability.
Graceful Shutdown for Node.js Services
Learn how to drain HTTP requests, stop background work, close dependencies, and make Node.js deployments terminate safely.
Optimistic Concurrency Control for APIs and Databases
Learn how to prevent lost updates with version columns, ETags, compare-and-swap writes, and useful conflict responses.
Practical API Rate Limiting with Token Buckets
Learn how to design token-bucket API rate limits that protect services without punishing normal users.
Zero-Downtime Database Migrations with Expand and Contract
Learn how to ship database schema changes safely with expand-contract migrations, batched backfills, compatible application deploys, and clear rollback points.
Implementing the Transactional Outbox Pattern for Reliable Events
Learn how the transactional outbox pattern keeps database writes and event publication consistent without distributed transactions.
Designing Dead-Letter Queues That Help You Recover Events
Learn how to design dead-letter queues with useful metadata, triage workflows, safe replay tools, and clear ownership so failed events can be recovered instead of ignored.
Implementing Idempotency Keys in APIs to Prevent Duplicate Actions
Learn how idempotency keys prevent duplicate side effects in retry-heavy clients by combining request fingerprinting, state tracking, and careful concurrency handling.
Building Resilient Distributed Systems with Chaos Engineering
Learn how to use Chaos Engineering to make your distributed systems more resilient and reliable.