System Design Roadmap 09 - Integration Review and Interview Drills

Design a URL shortener: core components

API service, key generator, URL database, cache for hot redirects, analytics/event queue, abuse prevention, and CDN/edge routing if global.

URL shortener: main bottleneck

Read-heavy redirects. Use cache/CDN, short key indexes, horizontal app scaling, and careful collision/key-generation strategy.

Design a news feed: core components

Post service, social graph/follow service, feed generation, ranking, cache, fan-out workers, notification/event queue, and media storage/CDN.

News feed fan-out tradeoff

Fan-out on write precomputes feeds for faster reads but can be expensive for celebrity users. Fan-out on read saves writes but makes reads slower.

Design chat: core components

Connection gateway/WebSockets, message service, message store, delivery receipts, presence, push notifications, queues, and per-conversation ordering.

Chat consistency concern

Users expect ordered messages within a conversation and no lost messages. Use sequence IDs, durable storage before ack, retries, and idempotent sends.

Design video/image sharing: core components

Upload service, object storage, metadata DB, processing pipeline, CDN, transcoding/thumbnail workers, moderation, and feed/search indexes.

Media pipeline pattern

Accept upload, store original, enqueue processing, generate derived assets, update metadata, and notify clients. Make each step idempotent.

Design payments: core components

Payment API, ledger, idempotency keys, external processor integration, reconciliation jobs, audit logs, fraud/risk checks, and strict authorization.

Payment design invariant

Never double-charge or lose money movement. Prefer durable ledger records, idempotent requests, strong consistency around balances, and reconciliation.

Design metrics/analytics: core components

Event ingestion, queue/stream, processors, time-series or columnar storage, aggregation jobs, dashboards, alerts, and retention policies.

Analytics consistency tradeoff

Analytics can often be eventually consistent. Prioritize high write throughput, cheap storage, deduplication, and late-arriving event handling.

Design maps/proximity search: core components

Geospatial index, location updates, query API, caching, routing service, map tiles/CDN, and privacy controls.

Geospatial indexing options

Geohash/S2/H3 partition space into cells for proximity queries. Choose cell precision based on density, query radius, and update frequency.

Design search: core components

Source database, indexing pipeline, search cluster, ranking, query service, autocomplete, cache, and reindexing/backfill workflow.

Search freshness tradeoff

Near-real-time indexing improves UX but costs more and complicates writes. Batch indexing is cheaper but returns stale results.

Design file storage: core components

Metadata service, object/blob storage, upload/download APIs, permissions, versioning, replication, CDN, and background virus/preview jobs.

System design closing checklist

Call out bottlenecks, failure modes, consistency choices, security, monitoring, scaling path, cost tradeoffs, and what you would test first.

Most common interview mistake

Jumping to components before requirements and scale. First clarify users, operations, data size, latency/availability targets, and constraints.