Notes / System Design / 15 Complete Case Studies / 01 Telemetry Ingestion Pipeline

3.4 Scaling Each Layer

Scaling unit and trigger for every layer of the telemetry ingestion pipeline, from the ingestion gateway through to storage.

Appears in: Telemetry Ingestion Pipeline — §3, Deep Dives — this is §3.4.

3.4 Scaling Each Layer

LayerScaling unitScaling triggerNotes
Ingestion gatewayPod replicas (HPA)CPU > 70% OR active connections > 10KStateless; L7 load balancer distributes
Kafka brokersPartition countConsumer lag growing + broker CPU > 80%Increase partitions; re-assign partition leaders
Metric processorConsumer group podsConsumer lag > target (e.g., 60s worth)Pods = partition count for maximum parallelism
Trace assemblerStateful shard podsMemory > 70% OR assembly latency risingResharding is costly; over-provision or use consistent hashing
Log processorConsumer group podsConsumer lagStateless; easy to scale
Storage (Mimir)Ingester podsSeries per ingester > targetCortex/Mimir uses ring-based consistent hashing for ingesters

Local graph

Full graph →

Linked from 7 notes

3 — Prometheus in the Observability Ecosystem

Where Prometheus sits in the CNCF landscape — its pull-based cloud-native origins, its companion projects, and where this book does (and doesn't yet) connect it to the wider stack.

2 — Long-Term Storage

Why single-node Prometheus has no built-in long-term-storage or HA story, and how remote_write receivers like Thanos, Cortex, Mimir, and VictoriaMetrics fill that gap at a horizontally-scaled, multi-tenant layer.

3.7 Data Tiering and Compaction (Mimir/Thanos)

Data tiering and compaction in Mimir/Thanos — the ingester-to-object-store journey, compaction levels, vertical compaction/dedup, compaction storms, and the config knobs that control them.

Q3: Trace Sampling Loses Spans During Incident Peaks — Redesign

Full principal-level solution: the trace pipeline drops spans exactly when incidents spike trace volume — diagnose the failure mode and redesign the tail-sampling pipeline to survive it.

Q6: Compactor Queue Backing Up During a Multi-Tenant Flush

Full principal-level solution: diagnose a compactor backlog causing query latency spikes during a large multi-tenant flush, and mitigate it without pausing ingestion.

Prometheus

A book-shaped table of contents for Prometheus: monitoring foundations through architecture, data model, instrumentation, service discovery, PromQL, alerting, production operation, PCA certification, and MAANG interview prep — cross-linking existing notes instead of duplicating them.

Chapter 1 — Telemetry Ingestion Pipeline

Principal/Staff-level design of a high-throughput telemetry ingestion pipeline — requirements, architecture, deep dives, and trade-offs at 10x scale.