システム設計面接キット

Reference latency and unit tables, assumptions register, QPS, storage, bandwidth, cache, server and cost formulas, worked example, sensitivity, traceability

Failure model, deadlines, retries with jitter and budgets, circuit breakers, bulkheads, backpressure, criticality-based shedding, degradation, fault tests

System design at scale – requirements, capacity estimates (KaTeX), API, data model, HLD, deep dives, reliability, cost, trade-offs

Capacity (KaTeX), API, base62 key generation and collisions, 301 vs 302, caching, click analytics, abuse protection, custom aliases, expiry, multi-region

Distributed rate limiter: token bucket vs window algorithms, atomic Redis scripts, multi-region shares, 429 headers, fail-open

Consistent hashing, vnodes, quorums (R+W>N), vector clocks vs LWW, hinted handoff, Merkle anti-entropy, gossip, LSM compaction, cache eviction, hot keys

Partitions, log segments, retention, acks and idempotence, consumer groups, offsets, delivery semantics, ordering, ISR replication, backpressure, DLQ, schemas

Social feed: hybrid fan-out on write and read, celebrity threshold, ranking pipeline, feed cache, snapshot pagination, media

Real-time 1:1 and group chat: WebSocket gateways, per-conversation ordering, fan-out, receipts, offline sync, push, E2EE

Push, email, SMS and in-app: templates, preferences and opt-out, priority queues, caps, provider failover, retries, dedup

Seat and room inventory, holds with TTL, overbooking math, conditional writes, idempotency, payment saga, waiting room, search vs booking paths

Typeahead: trie/FST/edge n-gram index, popularity and recency ranking, query-log pipeline, edge cache, latency budget, safety

Distributed scheduler: cron, delayed and one-off jobs, leases and visibility timeouts, fencing, idempotent handlers, retries, DLQ

Payments and double-entry ledger: idempotency keys, payment state machine, PSPs, outbox, reconciliation, audit, PCI DSS scope

Ride matching: location ingestion, H3/geohash index, batched matching, ETA, trip state machine, single assignment, surge note

Video upload and ABR streaming: chunked transcoding, renditions ladder, CMAF HLS/DASH, CDN, DRM note, view counts, storage tiers

Real-time co-editing: OT vs CRDT, sequence CRDT, state-vector sync, presence, offline merge, snapshots and compaction, access

Gateway, router, continuous batching, paged KV cache, GPU memory and throughput math, TTFT/TPOT SLOs, autoscaling, quantization, HNSW/IVF, cost per token

UUIDv4 and v7 per RFC 9562, Snowflake 41/10/12 layout, clock skew and backwards clocks, leased worker IDs, index locality, 53-bit JS limit, ticket servers

Introductory: sorted sets, time-encoded tie breaking within 53 bits, top N and my rank, sharded ranks, friends boards, seasonal resets, anti-cheat boundary

Distributed cache: cache-aside, delete on write, leases against herds, commit-log invalidation across regions, slot maps, hot keys, gutter pool, warm-up

Resumable direct upload, media pipeline, display sizes, storage tiers, 64-bit time-sortable ids, sharded PostgreSQL, celebrity fan-out, CDN

Inverted index, segments, real-time ingestion vs queries, document and time sharding, replicas, BM25 plus reranker, autocomplete boundary, relevance evaluation

Vector tiles and CDN, region-sharded routing graph with customizable contraction hierarchies, live-traffic ETA, rerouting, geocoding, offline packs