Flashcard Decks

Parallel scatter-gather, deadline propagation, hedged requests, and partial-failure policies.

Nine principal/staff-level scenarios on the fan-out/fan-in pattern — partitioning, hedging trade-offs, cancellation leaks, aggregator complexity, retry storms, backpressure, hierarchical fan-out, and pre-rollout validation.

Pattern-specific telemetry beyond RED metrics for fan-out/fan-in systems — width, per-shard latency, aggregation cost, partial results, cancellation, and hedging.

The sampling decision point determines whether a trace pipeline needs to buffer spans in memory — head-based decides at trace start, tail-based decides after the trace completes.

Three ways to enforce rate limits across a replicated gateway fleet — centralized Redis token bucket, decentralized gossip-based estimation, and Envoy's sidecar-plus-global-service pattern — and their precision/SPOF/latency trade-offs.

Every retry decision is made along three axes — trigger, backoff, budget — before delivery semantics even enter the picture. At-least-once, at-most-once, and exactly-once are the accumulated side effect of those decisions, not a separate design choice.

The ten protocol-specific gateways in the ingestion frontier — what each accepts, from whom, and when to use it.

High-level design of a MAANG-scale telemetry ingestion pipeline — layers, durability boundary, cardinality enforcement, temporality, multi-tenancy, data tiering, and global topology.

Twelve principal/staff-level system design scenarios on the telemetry ingestion pipeline — the core diagnosis, redesign, or trade-off decision behind each answer.