Networking & Protocols
Loss detection and congestion control
Crux QUIC uses monotonically increasing packet numbers and pluggable congestion control — NewReno, CUBIC, BBR — eliminating Karn''''s algorithm ambiguity and enabling per-application CC upgrades.
Your altitude — climbing toward senior
ZeroJuniorMiddleSenior
You are at middle altitude — in the skyTCP retransmits a dropped packet with the same sequence number. The ACK that comes back is ambiguous — did the server acknowledge the original send or the retransmit? QUIC gives every packet a unique number, and this single change makes loss detection unambiguous and congestion control pluggable.
Monotonic packet numbers
QUIC packet numbers are strictly increasing integers — a lost packet gets a new packet number on retransmit. Unlike TCP where retransmitted packets carry the same sequence number (creating ambiguity about which send the ACK acknowledges — Karn’s algorithm workaround), QUIC never reuses a packet number.
ACK ranges: QUIC ACK frames list ranges of acknowledged packet numbers:
ack_ranges: [1000..1005, 1008..1010]Gap at 1006 and 1007 means those packets are not yet received. This is equivalent to TCP SACK but native to the protocol.
Loss declaration happens when:
- A later packet is acknowledged and a reorder window has expired (typically 1.5× max RTT after the out-of-order packet was sent).
- Or the PTO (Probe Timeout) fires — analogous to TCP’s RTO but computed per packet-number space.
The monotonic model is simpler than TCP’s but requires retransmitted frames to be re-serialized with new packet numbers — a minor CPU cost.
Packet number model: TCP vs QUIC
TCP (ambiguous)
Send seq=100 → LOST
Retransmit seq=100 → ACK 101
ACK: original or retransmit?
→ Karn’s algorithm required
QUIC (unambiguous)
Send pkt#100 → LOST
Retransmit as pkt#101 → ACK 101
ACK#101 = retransmit confirmed
→ No ambiguity, no workaround
Pluggable congestion control
Because QUIC lives in user space, congestion control is pluggable. RFC 9002 specifies NewReno as the baseline. Major implementations ship:
- NewReno — slow start, additive increase, halve cwnd on loss.
- CUBIC (quiche from Cloudflare, Linux kernel standard) — aggressive probing after loss with a cubic function.
- BBR (Google, YouTube, Cloudflare edges) — model bottleneck bandwidth and minimum RTT; ignore random loss.
- HyStart++ — exit slow start early by detecting queueing, used with CUBIC.
Applications can select the algorithm at connection open. A single QUIC library can support multiple algorithms. This agility is impossible with kernel TCP where changing congestion control requires recompiling or sysctl changes across millions of servers.
The trade-off: Flexibility makes benchmarking hard — comparing two QUIC stacks with different CC algorithms is not a fair comparison of QUIC itself.
Quiz
Completed
Why does QUIC avoid Karn's algorithm, and what design choice makes this possible?
Heads-up QUIC does retransmit lost data but with new packet numbers. There is no FEC in the base spec.
Heads-up Karn's algorithm is about retransmit ambiguity, not wraparound. The key is unique packet numbers per send.
Heads-up UDP has no ACK mechanism at all. QUIC implements its own on top of UDP.
Quiz
Completed
Why is pluggable congestion control in QUIC significant compared to kernel TCP?
Heads-up UDP is just a transport layer; the CC algorithm is an application-level choice in QUIC regardless of UDP.
Heads-up The number of packet numbers is not the key differentiator. The key is user-space deployability.
Heads-up Speed is not the reason. User-space placement enables CC evolution without OS updates.
Which RFC?
Completed
Which RFC defines QUIC's loss detection algorithm and congestion control baseline?
Why this works
Why not just use TCP’s congestion control in QUIC? TCP’s congestion control is implemented in the kernel’s socket layer, where it has direct access to packet timestamps, kernel buffers, and NIC offload. Reimplementing it in user space (as QUIC does) costs CPU — each packet must be processed by user code without hardware assistance. The benefit: user space can iterate faster, can access application-level metadata (e.g., video stream priority) to make smarter CC decisions, and can be updated with the application without touching the OS.
- RFC 9002 baseline
- NewReno (slow start + AIMD)
- Cloudflare quiche default
- CUBIC (aggressive probing)
- Google YouTube / GCP
- BBR (bandwidth-RTT model)
- User-space CC update cycle
- weekly (vs monthly+ for kernel TCP)
- Per-connection CC selection
- at connection open
- 01What is Karn's algorithm and why does QUIC not need it?
- 02How does QUIC's ACK range mechanism work, and what TCP feature does it resemble?
- 03What is PTO in QUIC and how does it differ from TCP's RTO?
Recap
TCP’s sequence number reuse on retransmit creates Karn’s algorithm ambiguity — you cannot measure RTT from a retransmitted ACK. QUIC eliminates this by assigning every send a new, strictly increasing packet number. ACK frames carry ranges so gaps in acknowledged numbers reveal lost packets exactly, equivalent to TCP SACK but always active. Loss is declared via a reorder window or PTO (computed per packet-number space). Because QUIC runs in user space, congestion control is pluggable: RFC 9002 specifies NewReno as the baseline, but CUBIC and BBR ship in major libraries and can be selected per-connection. Applications on CDN edges can update their CC algorithm weekly without touching the kernel.
Connected lessons
deepens into
appears again in162
- The journey of a request: seven stops from socket to responsejunior
- Accept and parse: from kernel queue to a typed requestmiddle
- Routing and middleware: choosing what runs, and in what ordermiddle
- Handler and response: from business logic to bytes on the wiremiddle
- Streaming and backpressure: when the client reads slower than you writesenior
- Timeouts and tail latency: budgets, deadlines, and the fan-out trapsenior
- Middleware and DI: the two patterns that shape every backendjunior
- Writing middleware: signatures, next(), and the three framework modelsmiddle
- Inversion of control: how dependencies reach a classmiddle
- DI scopes and lifecycles: singleton, request, transientmiddle
- DI as a testing seam: fakes, mocks, and the boundary that matterssenior
- DI containers in production: resolution graphs, circular deps, and when not tosenior
- Blocking vs non-blocking I/O: two ways to waitjunior
- The event loop: one thread, ordered phasesmiddle
- What blocks the loop: CPU work and sync callsmiddle
- Offloading CPU work: worker threads and the libuv poolmiddle
- Backpressure and bounded concurrencysenior
- Throughput under load: tail latency and saturationsenior
- Why pool: the cost of creating a connectionjunior
- Pool sizing: why bigger is not fastermiddle
- Acquisition and timeouts: the wait queue is the real latency dialmiddle
- Retry strategies: backoff, jitter, and thundering herdmiddle
- Observability, production failures, and global-scale designsenior
- Tasks, microtasks, and scheduler.yield()middle
- Timer accuracy, throttling, and idle workmiddle
- Node.js event loop: phases, nextTick, and loop lagsenior
- Rendering strategies: SSG, SSR, ISR, streaming, and hydrationjunior
- SSG, SSR, ISR, streaming, and RSC — how each worksmiddle
- Hydration cost: selective, progressive, islands, resumabilitymiddle
- Core Web Vitals: what LCP, INP, and CLS measurejunior
- LCP: four phases, one dominant costmiddle
- INP: input delay, processing, presentationmiddle
- Lab vs field: why the two disagree and how to use eachmiddle
- Metric tradeoffs, RUM attribution, and the CI+field loopsenior
- The full picture: URL to LCP to INP as a relay racejunior
- Eight layers traced: from the service worker to the second navigationmiddle
- Five canonical breaks: where production reliably diessenior
- The three-track method: reading traces and building a monitored systemsenior
- What an index is and how it speeds up queriesjunior
- The leading-column rule and composite index designmiddle
- Partial, expression, and covering indexesmiddle
- Index types: GIN, GiST, BRIN, Hash, Bloom, and HOT updatesmiddle
- Index-only scans, the Visibility Map, and INCLUDEsenior
- Production failure modes and the index audit playbooksenior
- Index design exercise: full-text search strategysenior
- EXPLAIN and execution plans: what the planner decides and whyjunior
- Scan types: Seq, Index, Bitmap, Index-Onlymiddle
- Join algorithms and the row-estimate cascademiddle
- pg_statistic, ANALYZE, and production observabilitymiddle
- Extended statistics: fixing correlated-column estimate failuressenior
- Plan cache, cost-constant tuning, and planner internalssenior
- Production failure modes and plan stabilitysenior
- Connection pools: amortising the cost of a Postgres backendjunior
- PgBouncer session, transaction, and statement modesmiddle
- Pool sizing: the (cores × 2) + spindles formula and the two-layer stackmiddle
- Pool exhaustion and idle-in-transaction: the 3 AM failure modemiddle
- Migrating to transaction mode: rollout playbook and PgBouncer 1.21 prepared statementsmiddle
- The Postgres process model and why raising max_connections degrades throughputsenior
- Pooler landscape 2026, serverless connection storms, and the full failure-mode taxonomysenior
- ADD COLUMN: instant in PG 11+ vs rewrite in older Postgresjunior
- The lock-queue failure mode: why instant DDL can freeze the databasemiddle
- Safe DDL patterns: NOT VALID, CONCURRENTLY, and unsafe-op fixesmiddle
- Migration failure taxonomy and production disciplinesenior
- Shard-key selection: hash, range, list, and directory strategiesmiddle
- Co-location and Citus: the invariant that makes sharding usablemiddle
- The hot-shard failure mode: detection, isolation, and durable policymiddle
- Online resharding, 2PC, and the operational cost of shardingsenior
- The seven acts: from CREATE TABLE to Citusjunior
- Acts 1–3 in depth: schema, indexes, and planner statisticsmiddle
- Acts 4–6 in depth: MVCC bloat, connection pooling, and safe migrationsmiddle
- Act 7 in depth: sharding, co-location, and the seven-tier tradeoff cascademiddle
- Observability, anti-patterns, and production triagesenior
- What the three signals are: logs, metrics, and tracesjunior
- Metrics and cardinality: the cost model of a time-series databasemiddle
- Logs and volume: the cost model of structured loggingmiddle
- Traces and sampling: the cost model of distributed tracingmiddle
- Join keys and exemplars: making the three signals composemiddle
- Observability 2.0: wide events and the cost shiftsenior
- Failure modes and engineering practice: cardinality budgets, PII, and samplingsenior
- Why structured logs exist: the diary vs the spreadsheetjunior
- The production log schema: fields every line must carrymiddle
- Log levels and alert routingmiddle
- Sampling strategies and log costmiddle
- PII redaction and log injectionsenior
- Trace context propagation in logssenior
- OTel Logs Data Model and audit logs as a subsystemsenior
- OTel signals, Semantic Conventions, and the OTLP wire formatmiddle
- Auto-instrumentation and manual spans: the 80/20 of OTelmiddle
- The OTel Collector: receivers, processors, exporters, and deployment patternsmiddle
- Sampling strategies: head, tail, and parent-basedmiddle
- Vendor neutrality, eBPF instrumentation, the Operator, and browser/serverless OTelsenior
- Operating the OTel Collector: reliability, version skew, failure modes, and governancesenior
- RED and USE: two checklists, one triage disciplinejunior
- Instrumenting RED in Prometheus: counters, histograms, and cardinality disciplinemiddle
- USE on Linux: CPU, memory, disk, network, and PSImiddle
- Golden signals, dashboard layout, and service mesh auto-REDmiddle
- Cardinality as a cost driver: labels, PII, exemplars, and samplingmiddle
- Native histograms, SLO tie-in, and production failure patternsmiddle
- Choosing SLIs and SLO targets: ratios, not feelingsmiddle
- Multi-window multi-burn-rate alerting: why AND beats ORmiddle
- Error budget policy, latency SLOs, and composite journeysmiddle
- Iceberg SLIs, composite SLO math, and SLA vs SLOsenior
- Flame graphs: reading the picture that shows where time goesjunior
- Sampling vs instrumentation profiling: why 99 Hz wins in productionmiddle
- Profile types: CPU, memory, off-CPU, mutex — which one to reach formiddle
- Continuous profiling: always-on flame graphs with eBPF and trace-id correlationmiddle
- How flame graphs are built from samples, and the production workflows that use themmiddle
- Linux perf, eBPF internals, PGO, and the limits of samplingsenior
- Profiling in production: security, war stories, OTel profiles, and the infrastructure designsenior
- The debugging funnel: SLO → RED → trace → profilejunior
- OTel architecture: one SDK, four signals, one wire formatmiddle
- Cost discipline: keeping observability under 5% of infra spendmiddle
- Scale, security, and the ROI of observable systemssenior
- Why profile first: measure where time actually goesjunior
- Amdahl''''s law and self-time: the ceiling on every speedup you can shipmiddle
- The measurement loop: microbench, macrobench, prod profile, observer effectmiddle
- Reading flame graphs: shapes, per-language profilers, and the 60-second scanmiddle
- Statistical baselines: why one run is not a measurementmiddle
- Profiler history and microbenchmark pitfalls: Knuth to GWPsenior
- Hardware counters, cold-start profiles, and profile securitysenior
- Continuous profiling at scale: costs, CI gates, trace correlation, and anti-patternssenior
- What makes a hot path: symptom vs causejunior
- Five shapes of hotspot: CPU, alloc, cache, lock, syscallmiddle
- Reading parent and child chains: where to apply the fixmiddle
- JIT deopt, the fix-and-verify loop, and PR-time profilingmiddle
- Hardware counters and Intel TMA: sub-category diagnosissenior
- False sharing and native-bridge hot pathssenior
- Hot paths in production: security, tail latency, and tooling lineagesenior
- Memory hierarchy: why the same O(N) loop can be 17x slowerjunior
- Row-major vs column-major: access order and the 9x gapjunior
- Branch prediction and branchless codemiddle
- Hardware prefetcher, TLB, and memory-level parallelismsenior
- GC basics: what the runtime taxes you forjunior
- GC algorithms: generational, concurrent, and per-runtimemiddle
- GC tradeoffs: pause, throughput, heap — and object poolingmiddle
- GC tuning: pacing, heap shape, and allocation observabilitymiddle
- GC internals: tri-color invariant, write barriers, and per-runtime deep-divessenior
- GC in production: observability, security, edge cases, and fleet governancesenior
- N+1: one logical operation, many round-tripsjunior
- Fix families: JOIN, IN, preload, and DataLoadermiddle
- Detecting N+1: query logs, APM traces, and CI gatesmiddle
- DataLoader: batching across resolver treesmiddle
- Cross-protocol N+1: HTTP fan-out and Redis MGETmiddle
- N+1 at scale: pool exhaustion, plan changes, and denormalisationsenior
- Batching: amortize fixed cost per operationjunior
- The batching window: size and wait timemiddle
- Batching in Kafka and Postgresmiddle
- io_uring and observability of batchingmiddle
- From Nagle to io_uring: evolution of batchingmiddle
- Backpressure, failure isolation, and batch security in productionsenior
- What a bundle actually costs: download, parse, compile, executejunior
- Core Web Vitals: LCP, INP, and CLSmiddle
- Code splitting: route-level, component-level, vendor splittingmiddle
- Tree shaking and compression: removing what you don''''t usemiddle
- Third-party scripts: the silent budget killermiddle
- CI enforcement and RUM: making budgets stickmiddle
- V8 JIT pipeline, HTTP priorities, and bundle securitysenior
- The performance loop: discipline, not a projectjunior
- Classify and fix: matching bottleneck families to remediesmiddle
- Observability stack and CI gates: catching regressions before they shipmiddle
- Incident to enforcement: SLO burn to verified fix in 35 minutesmiddle
- Culture, economics, and org-scale performancesenior