Frontend Architecture
Senior internals: RSC payload, caching layers, and production failure modes
After an RSC migration at Vercel, a customer’s TTFB regressed from 100ms to 2.5s. The component tree looked correct. The data fetching looked correct. The culprit: a Server Component that fetched 50 items in a sequential loop instead of Promise.all. A pattern that is invisible in code review but catastrophic in production.
The RSC Payload wire format
When React renders a Server Component tree, the output is not HTML and not JSON. It is the RSC Payload — a custom streaming format designed to be processed incrementally as it arrives over the network.
The format uses row-prefix codes:
M— module reference (client component bundle URL)J— JSON literal (serialized props or data)H— hint (browser directive likeprefetch)
Each row references previous rows by ID, building up the React tree incrementally. Example:
0:["$","div",null,{"children":[["$","h1",null,{"children":"Product"}]]}]
1:{"id":42,"name":"Product","price":29.99}The browser’s React client processes these rows as they arrive, enabling the same “stream HTML, render what you have” behaviour described in the Suspense streaming lesson — but for React tree updates, not just initial HTML.
The RSC Payload is the mechanism behind Next.js client-side navigation: instead of fetching a new HTML page, the client fetches the RSC Payload for the new route and applies it to the existing React tree. This makes navigations feel instant because the shell (layout, nav) never re-renders.
Server Functions: type-safe RPC over HTTP
React 19 introduces Server Functions — async functions marked with 'use server' that can be called from Client Components but execute on the server. The framework generates a stable POST endpoint for each; the client-side call becomes a typed fetch.
// actions.ts
'use server';
export async function likePost(postId: string) {
await db.post.update({ where: { id: postId }, data: { likes: { increment: 1 } } });
revalidatePath('/posts');
}
// PostCard.tsx — Client Component
'use client';
import { likePost } from './actions';
export function PostCard({ postId }) {
return <button onClick={() => likePost(postId)}>Like</button>;
}TypeScript sees likePost as a normal typed function. No need to define a REST route, request/response types, or error handling shape. Arguments are serialized automatically; the function body never ships to the client.
Combined with form actions (<form action={serverFunc}>), Server Functions enable progressive enhancement: the form works with JavaScript disabled because the browser submits to the generated endpoint directly.
Constraints: arguments must be serializable (no class instances, no functions). Authentication must be explicit inside the function — the framework does not automatically check session cookies. Never trust client-provided IDs for ownership checks without re-validating server-side.
Caching layers and coordinated invalidation
A modern Next.js 15 app has at least five caching layers:
- CDN edge cache — URL-keyed, seconds to hours. Serves static content without hitting origin.
- Next.js fetch cache — RSC
fetch()calls are cached by default with a configurablerevalidateduration. - Database query cache — Redis or in-memory, sits in front of the database.
- Browser HTTP cache — controlled by Cache-Control headers.
- Client library cache — TanStack Query / SWR, keyed by queryKey.
A mutation that changes data must invalidate the right layers:
// On the server (Server Function or route handler)
revalidatePath('/products'); // Next.js cache + CDN if configured
revalidateTag('product-42'); // granular tag-based CDN purge
// On the client
queryClient.invalidateQueries({ queryKey: ['product', 42] });Skipping any layer leaves users staring at stale data. The most common bug: the server cache is revalidated but the client TanStack Query cache still serves the old response for the next 5 minutes.
Hydration mismatches: silent interactivity bugs
React’s hydration assumes the server-rendered DOM exactly matches what the client would render given the same props. When they differ, React’s reconciler tries to patch the DOM, but its model of which event handlers belong to which DOM nodes is now inconsistent — clicks may fire on the wrong handler or be silently dropped.
Common causes:
Date.now()orMath.random()rendered on both sides — different values- Conditional rendering based on
window.innerWidth— server has nowindow - Reading
localStorageat module top-level
The visible output looks correct because both server and client render valid HTML. But the event-handler bindings are misaligned.
Fix patterns:
- useEffect for client-only logic: render the safe initial state on both sides; mutate to client-specific state after hydration completes
- suppressHydrationWarning: valid for genuinely divergent leaves (timestamps with locale)
- Client Components for content that is fundamentally client-specific
React 19 improved hydration error messages to localise the mismatched element; pre-19, the message was generic. Production teams instrument hydration errors via error boundaries reporting to Sentry — a steady non-zero rate signals accumulating mismatches before they surface as user-reported bugs.
Edge runtime for data fetching
The edge runtime (Next.js Edge, Cloudflare Workers, Vercel Edge) runs server code at CDN nodes instead of a central data center. Latency to the user: single-digit ms vs hundreds for traditional servers.
Best fit: globally-replicated data sources (Cloudflare D1, Turso, PlanetScale) where the edge server fetches from a nearby replica. Edge-rendered RSC gives the user first paint in 50–100ms regardless of geography.
Constraints: limited APIs (no Node.js fs, no native modules), Web-standard APIs only (fetch, crypto.subtle), smaller memory limits per request.
Why this works
The RSC Payload format is documented in React’s RFC repository and is intentionally not JSON — it is designed for incremental parsing as bytes arrive. JSON requires the full string before parsing. The RSC format allows React to begin constructing the component tree from the first bytes, matching the streaming model of chunked HTTP transfer. Framework authors (Next.js, Remix, Waku) implement the server emitter; React DOM implements the client consumer.
Real production failures
Vercel 2024 — server-side waterfall: A customer’s RSC migration regressed TTFB from 100ms to 2.5s. Root cause: a Server Component fetched 50 product items in a sequential for...await loop instead of Promise.all. Caught by observability dashboards flagging the p95 TTFB spike; fixed by wrapping in Promise.all. The lesson: sequential await in Server Components is a server-side waterfall with the same cost as client-side — it blocks the Suspense boundary from streaming.
Spotify Web Player 2023 — incomplete optimistic snapshot: An optimistic update for “add to playlist” used setQueryData but the onMutate did not snapshot the previous state. When the server rejected the request, the rollback set the cache value to undefined instead of the actual previous state — playlists appeared to vanish for a few seconds. Fixed by the canonical pattern: always call queryClient.getQueryData in onMutate before the optimistic update.
Linear 2024 — SSE reconnect message loss: A Server-Sent Events based real-time sync used a client-generated message ID scheme. On reconnect, the client missed messages sent during the disconnect window — users saw cards “bounce back” after drag operations. Fixed by using server-assigned monotonic IDs and a Last-Event-ID-based resume protocol that the server replays missed messages for.
- Edge runtime cold start
- under 50 ms
- RSC Payload size (typical page)
- 10–50 KB
- HTTP/2 max concurrent streams (default)
- 100
- Next.js 15 App Router default
- RSC + App Router
- React 19 RSC stable release
- Q4 2024
A team migrates from CSR to RSC. Bundle shrinks from 240KB to 60KB but JS evaluation time on mobile stays the same. Why?
Why do hydration mismatches silently break interactivity even when the visible HTML looks correct?
- 01What is the RSC Payload and why is it not JSON?
- 02Explain the 5 caching layers in a Next.js 15 app and which API handles each.
- 03What is the Spotify Web Player optimistic update bug and what is the fix?
The RSC Payload is a streaming row-protocol — not HTML, not JSON — that lets React incrementally build the component tree as bytes arrive, enabling navigation without full page reloads. Server Functions generate stable typed POST endpoints from 'use server'-marked async functions, replacing hand-authored API routes. Multi-layer cache invalidation spans five layers — CDN, Next.js fetch cache, DB query cache, HTTP cache, and client library cache — each needing explicit invalidation on mutations. Hydration mismatches corrupt event-handler bindings silently; fix by deferring client-only logic to useEffect and monitoring hydration errors in production. Edge runtime cuts first-paint latency to 50ms but only when the data source is globally replicated — edge functions calling a single-region database add round-trips, not subtract them.
appears again in202
- Why GraphQL gets N+1junior
- DataLoader mechanics: tick-boundary batchingmiddle
- Batch function contracts: ordering, shapes, errorsmiddle
- Federation and lookahead: batching beyond DataLoadermiddle
- Query complexity defences: depth, cost, persisted queriesmiddle
- Senior GraphQL API: scheduling contract, tenant isolation, observabilitysenior
- Why idempotency: making retries safejunior
- Server-side state machine: four states of an idempotency keymiddle
- Outbox and inbox: effectively-once across the dual-write boundarymiddle
- Concurrency and cache architecture for idempotency at scalesenior
- Observability, production failures, and global-scale designsenior
- The event loop: one thread, three queuesjunior
- Tasks, microtasks, and scheduler.yield()middle
- Microtask starvation, Long Tasks, and LoAFsenior
- Node.js event loop: phases, nextTick, and loop lagsenior
- React, Vue, and INP observability in productionsenior
- The render pipeline: six stages from bytes to pixelsjunior
- Stage costs and the renderer process modelmiddle
- Invalidation, dirty bits, and containmiddle
- Compositor layers: promotion, overlap, and GPU memorymiddle
- DevTools flame strip and the frame lifecyclemiddle
- Layout thrash: forced synchronous layoutsenior
- BeginMainFrame, compositor-driven animations, and GPU memorysenior
- Production observability: LoAF, INP, and the full attack surfacesenior
- What V8 is and why performance varies 100×junior
- V8''''s four-tier JIT pipeline and profile-guided tieringmiddle
- Hidden classes, transition trees, and memory layoutmiddle
- Inline caches, IC states, and deoptimizationmiddle
- Orinoco GC: parallel scavenger, concurrent marking, and write barriersmiddle
- TurboFan''''s speculative engine and the deopt-loop trapsenior
- V8 in production: isolates, pointer compression, and real failuressenior
- What workers are and why they existjunior
- Web worker mechanics: dedicated, shared, and OffscreenCanvasmiddle
- Structured clone and transferablesmiddle
- Service worker lifecycle and cache strategiesmiddle
- SharedArrayBuffer, Atomics, and cross-origin isolationsenior
- Service worker edge cases: version skew, durability, and navigation trapssenior
- Worker pools, Comlink, and production observabilitysenior
- What the reconciler does: render vs commitjunior
- The fiber object and the double-buffer treemiddle
- Render phase purity and commit phase sub-stepsmiddle
- Reconciliation: diffing heuristics and the key trapmiddle
- Priority lanes, time-slicing, and useTransitionmiddle
- Bailout, memoisation, and tearingsenior
- React Profiler, the Compiler, and production observabilitysenior
- Rendering strategies: SSG, SSR, ISR, streaming, and hydrationjunior
- SSG, SSR, ISR, streaming, and RSC — how each worksmiddle
- Hydration cost: selective, progressive, islands, resumabilitymiddle
- Hydration mismatch: causes, detection, and the determinism rulesenior
- RSC, per-route strategy, and production observabilitysenior
- Core Web Vitals: what LCP, INP, and CLS measurejunior
- CLS: why layout shifts happen and how to stop themmiddle
- 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 is a cache stampede and why it makes things worsejunior
- Lock and single-flight: bounding concurrent rebuildsmiddle
- XFetch: coordination-free probabilistic early expirationmiddle
- Stale-while-revalidate and CDN request coalescingmiddle
- Detecting stampedes and designing TTL for productionmiddle
- Metastable failure, fencing tokens, and production postmortemssenior
- What a relation is: tables, rows, keys, and constraintsjunior
- Constraints, keys, and Postgres data typesmiddle
- Normal forms, denormalization, and why schemas stickmiddle
- JSONB, arrays, and when a side table winsmiddle
- Heap storage, TOAST, and column alignmentsenior
- Schema integrity: deferral, versioning, and production failure modessenior
- Relational vs document, wide-column, graph, and key-valuesenior
- Index-only scans, the Visibility Map, and INCLUDEsenior
- Production failure modes and the index audit playbooksenior
- pg_statistic, ANALYZE, and production observabilitymiddle
- Production failure modes and plan stabilitysenior
- MVCC: why readers and writers never wait for each otherjunior
- Row versions and snapshots: the on-disk mechanicsmiddle
- HOT updates and isolation levels: what you gain and what you paymiddle
- Vacuum and bloat: keeping the storage tax boundedmiddle
- CLOG, XID wraparound, and MultiXact: deep visibility internalssenior
- SSI internals and production autovacuum tuningsenior
- Real-world MVCC failures, deployment patterns, and distributed snapshotssenior
- 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
- What a schema migration is and why it replaces ad-hoc DDLjunior
- 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
- Expand-contract: zero-downtime for breaking schema changesmiddle
- Advisory locks, migration tools, and deploy coordinationsenior
- Migration failure taxonomy and production disciplinesenior
- Why sharding exists: the single-Postgres ceilingjunior
- Shard-key selection: hash, range, list, and directory strategiesmiddle
- Partitioning vs sharding: same word, two different thingsmiddle
- Co-location and Citus: the invariant that makes sharding usablemiddle
- The hot-shard failure mode: detection, isolation, and durable policymiddle
- Schema-based sharding and multi-tenancy alternativessenior
- 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
- Raft roles, terms, and why majority quorums prevent split brainjunior
- How Raft replicates a log entry and decides it is safe to commitmiddle
- Raft leader election: timeouts, voting rules, and the four safety propertiesmiddle
- Raft in the real world: partitions, slow disks, and client routingmiddle
- Raft extensions: pre-vote, learners, snapshots, and linearizable readssenior
- Raft in production: membership changes, Multi-Raft, and observabilitysenior
- The IP envelopejunior
- Reading the IP headermiddle
- The three-way handshakejunior
- Sequence numbers and connection statemiddle
- DNS: what it does and why it existsjunior
- The resolver walk: referrals, record types, and gluemiddle
- TTL, caching, and DNS propagationmiddle
- What TLS does and why it existsjunior
- The 1-RTT handshake: key shares and ECDHEmiddle
- Session resumption and 0-RTTmiddle
- Key schedule, SNI, ALPN, and extensionssenior
- 0-RTT defenses, ECH, hybrid PQ, and production TLSsenior
- WebSocket: the HTTP upgrade handshakejunior
- WebSocket frame format: opcodes, masking, fragmentationmiddle
- WebSocket backpressure: when clients can''''t keep upmiddle
- Reconnection: jittered backoff, thundering herd, message resumptionsenior
- WebSocket at scale: HTTP/2 multiplexing, permessage-deflate, C10Msenior
- WebSocket in production: proxies, security, and distributed architecturesenior
- What reverse proxies dojunior
- Health checks, connection draining, and slow startmiddle
- Session affinity, consistent hashing, and the right fixmiddle
- Retry storms, circuit breakers, and load sheddingsenior
- Resilient LB architecture: anycast, zone-aware routing, and observabilitysenior
- Why QUIC and not TCP+TLSjunior
- Connection IDs and network migrationmiddle
- 0-RTT resumption and packet encryptionsenior
- DDoS: what it is and why it worksjunior
- Amplification attacks and state exhaustionmiddle
- Rate limiting: algorithms and architecturemiddle
- WAFs, firewalls, mTLS, and HSTSmiddle
- DNS cache poisoning and BGP hijackingsenior
- Defense-in-depth architecture and attack economicssenior
- The twelve layers: one URL, seven actorsjunior
- DNS, TCP, TLS in sequence: where the milliseconds gomiddle
- Proxy intercepts and security gates: rate limiters, WAF, mTLSmiddle
- Alternate paths: QUIC 0-RTT, WebSocket upgrade, connection migrationmiddle
- Observability: distributed traces, USE/RED, and samplingsenior
- Resilience: cascading retries, circuit breakers, and error budgetssenior
- What the three signals are: logs, metrics, and tracesjunior
- Why structured logs exist: the diary vs the spreadsheetjunior
- The production log schema: fields every line must carrymiddle
- PII redaction and log injectionsenior
- OTel Logs Data Model and audit logs as a subsystemsenior
- What is OpenTelemetry: API, SDK, Collector, OTLPjunior
- OTel signals, Semantic Conventions, and the OTLP wire formatmiddle
- The OTel Collector: receivers, processors, exporters, and deployment patternsmiddle
- Vendor neutrality, eBPF instrumentation, the Operator, and browser/serverless OTelsenior
- Operating the OTel Collector: reliability, version skew, failure modes, and governancesenior
- SLI, SLO, and the error budget: reliability by the numbersjunior
- Error budget policy, latency SLOs, and composite journeysmiddle
- Production SLO failures, self-observability, security, and the big picturesenior
- What is trace propagation and why broken propagation is worse than nonejunior
- traceparent and tracestate: the W3C header format in fullmiddle
- Baggage and async boundaries: carrying context across queues and callbacksmiddle
- Async context per language, service mesh, B3 migration, and securitysenior
- Production propagation failures, span links, and platform designsenior
- The debugging funnel: SLO → RED → trace → profilejunior
- OTel architecture: one SDK, four signals, one wire formatmiddle
- The incident loop: from pager to postmortem to preventionmiddle
- Scale, security, and the ROI of observable systemssenior
- Cache lines, struct layout, and false sharingmiddle
- SIMD, SoA vs AoS, and memory bandwidthmiddle
- Cache-oblivious algorithms, PGO, and production failuressenior
- GC in production: observability, security, edge cases, and fleet governancesenior
- 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
- 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
- At-most-once, at-least-once, exactly-once: the three delivery contractsjunior
- The three failure legs — where duplicates and losses actually happenmiddle
- Consumer-side dedup: the cheapest path to exactly-once processingmiddle
- Kafka exactly-once semantics: idempotent producer and transactionsmiddle
- SQS visibility timeout, DLQ, and the outbox patternmiddle
- Exactly-once in production: impossibility proof, hybrid patterns, and real incidentssenior
- What OAuth is and why passwords are not the answerjunior
- Authorization code flow with PKCEmiddle
- ID token validation and JWKS cache managementmiddle
- Refresh token rotation and scope-based least privilegemiddle
- Sender-constrained tokens: DPoP and mTLSsenior
- OAuth in production: audience attacks, observability, and real failuressenior