Glossary · 40 terms
Reliability & scale
The engineering patterns behind messaging that never double-sends and never drops: idempotency, retries and backoff, queues and dead-letter queues, backpressure, observability and the service-level objectives that hold it together.
Foundation· 5
Start here — the concepts everything else builds on.
- High AvailabilityHADesigning a service to stay up through component failures, using redundancy and no single points of failure, measured as an uptime target such as 99.9%.Read definition
- Horizontal ScalingHandling more load by adding more instances of a service rather than a bigger machine, which works best when each instance is stateless and interchangeable.Read definition
- Load BalancingSpreading incoming webhook and API requests across several healthy server instances so that none is overloaded and a failed one stops receiving traffic.Read definition
- Message QueueA durable buffer between your application and the WhatsApp Cloud API that absorbs send spikes, paces delivery to your throughput limit and survives restarts.Read definition
- ObservabilityThe ability to explain any behaviour of a messaging system, such as a sudden drop in delivery rate, from the logs, metrics and traces it emits.Read definition
Practitioner· 29
What you need once you are running it day to day.
- AlertingAlerting turns metrics and SLO burn rates into pages and tickets that fire when customers are affected, routed to someone who can act on them.Read definition
- BackpressureBackpressure is flow control in which a slow downstream stage makes upstream producers slow down, so queues and memory stay bounded.Read definition
- Canary ReleaseA canary release sends a small slice of live traffic to a new version, compares its errors and latency with the stable one, and rolls back if it degrades.Read definition
- Circuit BreakerA circuit breaker stops calling a dependency that keeps failing, fails fast for a cool-down period, then lets trial calls through to test recovery.Read definition
- Consumer LagConsumer lag is the backlog between messages produced and messages processed, measured as queue depth or as the age of the oldest waiting message.Read definition
- Dead-Letter QueueDLQA dead-letter queue (DLQ) holds messages that exhausted their retries or failed validation, so they can be inspected, fixed and redriven later.Read definition
- Disaster RecoveryDRDisaster recovery (DR) is the plan, backups and drills for restoring service after losing a region or data store, within agreed RPO and RTO targets.Read definition
- Distributed TracingFollowing one request or message across every service it touches, as timed spans that share a trace ID, to see where latency and errors build up.Read definition
- Error BudgetThe amount of failure an SLO allows over its window (100% minus the target), spent by incidents and used to decide when to slow releases down.Read definition
- Event-Driven ArchitectureEDAA design style where services publish events (message received, order placed) to a broker and others react asynchronously, instead of chaining calls.Read definition
- Exponential BackoffA retry schedule that multiplies the wait after each failure (1 s, 2 s, 4 s, 8 s…) up to a cap, giving a rate-limited or struggling service time to recover.Read definition
- FailoverAutomatically moving traffic or a role, such as the primary database, from a failed component to a standby so the service keeps running.Read definition
- Four Golden SignalsLatency, traffic, errors and saturation: the four measurements SRE practice recommends watching first for any user-facing service.Read definition
- Graceful DegradationKeeping the core customer journey working with fewer features when a dependency fails, instead of failing outright: a text menu when a Flow is down.Read definition
- Health CheckAn endpoint or probe that tells a load balancer or orchestrator whether an instance is alive and ready for traffic, so it can restart or route around it.Read definition
- Idempotency KeyA unique key created once per send request so that a retry after a timeout returns the first result instead of messaging the customer twice.Read definition
- Incident ManagementThe process a team follows to detect, triage, fix and learn from a live outage, with severity levels, named roles, customer updates and blameless postmortems.Read definition
- Load TestingSimulating broadcast-scale traffic against your send pipeline and webhook receivers, using a Cloud API mock, to find the breaking point before a campaign does.Read definition
- Mean Time to RecoveryMTTRThe average time from the start of an incident to service being restored for customers: a measure of how fast a team detects, responds and recovers.Read definition
- OpenTelemetryOTelA vendor-neutral open-source standard (APIs, SDKs and the OTLP protocol) for producing traces, metrics and logs and passing trace context between services.Read definition
- P99 LatencyThe time within which 99% of requests complete: a tail-latency measure that exposes the slow sends and webhook acknowledgements an average hides.Read definition
- Poison MessageA queued message or job that fails its consumer every time it is processed, so it retries forever unless it is detected and quarantined.Read definition
- Request TimeoutThe longest a client waits for a connection or a response before abandoning an API call, so a hung request fails fast instead of blocking.Read definition
- Retry PolicyThe rules that decide which failed API calls to try again, how many times and within what budget, and which errors must never be retried.Read definition
- Service Level IndicatorSLIA measured ratio of good events to valid events, such as sends accepted within one second, that shows how reliable a service feels to its users.Read definition
- Service Level ObjectiveSLOAn internal reliability target set on a measured indicator, such as 99.5% of OTP sends accepted within two seconds over 28 days.Read definition
- Structured LoggingWriting logs as machine-readable key-value records, usually JSON, so every event can be searched, correlated and alerted on by field.Read definition
- Synthetic MonitoringScheduled robot checks that run a real user journey, such as sending a test WhatsApp message and awaiting its status, to catch outages early.Read definition
- Token Bucket AlgorithmA rate-limiting algorithm in which tokens refill at a fixed rate up to a capacity and each request spends one, allowing short bursts but a steady average.Read definition
Expert· 6
Edge cases, internals and the details that matter at scale.
- At-Least-Once DeliveryA delivery guarantee under which every message reaches its consumer at least once — never lost, but possibly duplicated — so consumers must be idempotent.Read definition
- Backoff JitterBackoff jitter adds randomness to retry delays so clients that failed together do not retry together, spreading the load on a recovering API.Read definition
- Bulkhead PatternThe bulkhead pattern splits capacity into isolated pools — per tenant, number or traffic type — so one overloaded part cannot starve the rest.Read definition
- Load SheddingDeliberately rejecting or deferring low-priority traffic when a system is saturated, so critical messages such as OTPs and order updates still get through.Read definition
- Thundering HerdA failure mode in which many clients or workers hit a recovering service at the same instant, overwhelming it again just as it comes back.Read definition
- Transactional OutboxA pattern that saves an outgoing message in the same database transaction as the business change, then relays it, so neither happens without the other.Read definition
Put these terms to work.
Go live on the official WhatsApp Business API in 2 hours with a 7-day free trial — no card required.
7-day free trial Enterprise-grade reliability Live in 2 hours Built for Indian businesses