Fast properties, slow properties, and slack tracking
Three storage modes: in-object properties (inline in the body, single load), out-of-object fast properties (PropertyArray, +1 dereference), and dictionary mode (NameDictionary hashmap, ~50-100 cycles, set by deletes or too many props).
You construct a class with three fields, but the very first instances V8 allocates are bigger than they need to be — padded with empty slots. Then, after a few hundred instances, V8 quietly shrinks the class and frees the padding. This is slack tracking, and if you never knew it existed you have been mis-reading heap snapshots taken during warm-up. Property storage in V8 has three modes and one self-tuning allocator, and which one your object lands in decides whether a read is one instruction or a hash probe.
Three storage modes, three speeds
Why does storage location matter so much? Because a 50-100x cost difference between modes is the kind of cliff that shows up as a flat-line CPU flame chart but disappears in micro-allocations — you only find it by knowing what to look for before profiling.
The DescriptorArray from lesson 01 tells V8 where a field lives. There are three possible answers, in descending speed order:
- In-object properties. Stored inline in the object’s own memory block, right after the header. Reading one is a single load at a constant offset — the fastest possible. The number of in-object slots is fixed when the Map is created.
- Out-of-object (“normal”) fast properties. When an object outgrows its in-object slots, further properties spill into a separate PropertyArray that the object points to. These are still fast (fixed-offset, IC-friendly) but cost one extra pointer dereference: load the PropertyArray pointer, then load the slot. The DescriptorArray records a negative-ish index meaning “this field is in the PropertyArray at index N.”
- Dictionary mode (slow properties). When an object becomes too dynamic — too many properties, or any
delete— V8 abandons the fixed-offset layout entirely and switches the object to a NameDictionary: a per-object hash table from property name to value+attributes. Every access is now a generic hash lookup, roughly 50–100 cycles versus ~1 for an in-object load, and the inline cache cannot help.%HasFastProperties(obj)returnsfalse.
// d8 --allow-natives-syntax
const a = { p: 1, q: 2, r: 3 };
%HasFastProperties(a); // true — in-object, fixed offsets
delete a.q;
%HasFastProperties(a); // false — now a NameDictionary, permanently slowSlack tracking: V8 right-sizes your objects
Here is the self-tuning part the overview skipped. When you define a constructor or class, V8 does not yet know how many properties instances will end up with — this.x = ...; this.y = ... might be followed by more assignments in methods called later. Allocating exactly two in-object slots up front would force a PropertyArray spill the moment a third field appears.
So V8 over-allocates: the initial Map for the constructor reserves extra in-object slots (slack) — more than the constructor’s body declares. As instances are created and run, V8 watches the maximum number of in-object properties they actually use. After enough instances have been allocated (a fixed construction-count threshold), it completes slack tracking: it trims the unused slots, shrinks the instance size in the Map, and finalizes the Map. From then on, every new instance is allocated at the trimmed, exact size.
The consequences for a senior:
- Warm-up instances are larger than steady-state instances. A heap snapshot taken before slack tracking completes overstates per-object size. Measure after warm-up.
- The Map is not “final” immediately. Code that depends on a stabilized shape (and TurboFan’s specialization) only gets it after slack tracking finishes. Microbenchmarks that allocate a handful of objects can see a different shape than production at scale.
- Adding properties outside the constructor after finalization can no longer fit the in-object area and spills to a PropertyArray (or, if dynamic enough, dictionary mode).
What pushes an object into dictionary mode
Dictionary mode is the cliff. You fall off it via:
delete obj.propon a fast object — the single most common cause. V8 cannot leave a hole in a fixed-offset layout, so it converts the whole object to a NameDictionary. This is permanent: the object never climbs back to fast properties.- Too many properties added dynamically (the threshold is large, on the order of many hundreds to ~1000+, and depends on the version), so it is rarely the real culprit —
deleteis. - Adding properties in a way that defeats the transition tree — extreme shape churn can cause V8 to give up and switch to dictionary mode to stop allocating Maps.
Note const folding interacts here: when a field is const (only one value ever assigned across instances), V8 can store the value in the Map (in the descriptor) rather than in the object, so reads become “load the constant from the Map” — but the first divergent assignment loses constness (lesson 02) and the value moves into a real field. Dictionary mode abandons all of this.
- In-object read
- 1 load (~1 cycle)
- PropertyArray read
- 2 loads (+1 dereference)
- Dictionary-mode read
- ~50-100 cycles, hash probe
- What triggers dictionary mode
- delete, or extreme dynamism
- Recovery from dictionary mode
- none — needs a fresh object
- Slack tracking completes after
- a fixed construction-count threshold
- Detect mode
- %HasFastProperties(obj) in d8
A property read compiles to 'load PropertyArray pointer, then load slot 2 of that array'. Which storage mode is this property in?
A heap snapshot taken after allocating 10 instances of a class shows each instance is larger than a snapshot taken after allocating 10000. Why?
Order the slack-tracking lifecycle for a constructor's instances.
- 1 The initial Map reserves extra in-object slots beyond what the constructor body declares
- 2 Instances are allocated; V8 records the maximum in-object properties actually used
- 3 A fixed construction-count threshold of instances is reached
- 4 V8 trims the unused in-object slots and shrinks the instance size
- 5 The Map is finalized; every later instance is allocated at the trimmed exact size
▸Common mistake
A common self-inflicted dictionary-mode wound: using a plain object as a hash map with user-controlled keys (cache[userId] = entry) and then delete cache[userId] to evict. The first delete converts cache to dictionary mode — which is actually correct for a real map, but you also lose any IC benefit on it and pay hash-probe cost on every access. If the object is conceptually a dictionary, use a real Map from the start: it is built for arbitrary keys and deletions, never churns hidden classes, and has predictable performance.
- 01Describe the three property-storage modes and the cost of a read in each.
- 02What is slack tracking and why must you measure object size after warm-up?
- 03Why is `delete` worse than assigning null, in storage terms?
V8 stores an object’s named properties in one of three modes. In-object properties sit inline in the object’s body and read in a single load at a constant offset. When an object outgrows its in-object slots, extra properties spill into an out-of-object PropertyArray, costing one additional pointer dereference per read but staying fixed-offset and IC-friendly. Dictionary mode is the cliff: triggered chiefly by delete (and by extreme dynamism), it swaps the fixed layout for a per-object NameDictionary hash table, making every access a ~50-100 cycle hash probe with no inline-cache help, and it is permanent — only a fresh object recovers fast properties. Slack tracking is V8’s self-tuning allocator for fast objects: the constructor’s initial Map reserves extra in-object slots, V8 observes how many instances actually use as they are created, and after a fixed construction-count threshold it trims the slack, shrinks the instance size, and finalizes the Map — so per-object memory only settles after warm-up. Detect the mode with %HasFastProperties(obj), prefer null over delete, and reach for a real Map when keys are genuinely dynamic. Now when you see %HasFastProperties return false on a hot object, or a heap snapshot where object sizes shrink after warm-up, you know exactly which layer to investigate first.
Practice
Start at the top. Tasks go easiest → hardest: recall a fact, apply it to a case, then a senior-level stretch. Open one, attempt it, then reveal.
appears again in184
- 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
- Service worker lifecycle and cache strategiesmiddle
- Service worker edge cases: version skew, durability, and navigation trapssenior
- 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
- Where data fetching happens — and why it decides LCPjunior
- Fetch waterfalls — diagnosis and the Promise.all curemiddle
- React Server Components and Suspense streamingmiddle
- Client-side cache: TanStack Query, SWR, and stale-while-revalidatemiddle
- LCP, prefetch, and race conditions in interactive fetchingmiddle
- Senior internals: RSC payload, caching layers, and production failure modessenior
- 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
- The 1-RTT handshake: key shares and ECDHEmiddle
- Session resumption and 0-RTTmiddle
- 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
- 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
- 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
- The incident loop: from pager to postmortem to preventionmiddle
- 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
Something unclear?
Ask a question about this lesson. Questions are anonymous and go straight to the author to make the lesson better.