Skip to main content

Redis (Upstash)

Redis is an in-memory key-value store. It holds data in RAM, making reads and writes orders of magnitude faster than disk-based databases.

On this page

What is it?

Redis is an in-memory key-value store. It holds data in RAM, making reads and writes orders of magnitude faster than disk-based databases. Upstash is the managed provider — it exposes Redis over HTTP REST instead of a persistent TCP connection. The client is ~15KB and requires no connection pooling.

What is it for?

  • Rate limiting (sliding window counters per user/IP)
  • Session data (short-lived, fast-access tokens)
  • Feature flags (per-user or global toggles)
  • Leaderboards (sorted sets with automatic ranking)
  • Pub/sub messaging (real-time event broadcasting)
  • Caching hot queries (pre-computed results for expensive reads)
  • Atomic counters (view counts, download tallies)

Rule of thumb: Use Redis when data is ephemeral, needs sub-10ms access, or its disappearance is acceptable.

Why was it chosen?

Approach Serverless-safe Sub-10ms Rate limiting Free tier
In-memory Map No (resets per request) Yes No N/A
PostgreSQL UNLOGGED Yes No (~5–20ms) Manual Yes
Vercel KV Yes Yes No 30K cmd/month
ioredis (TCP) No (persistent conn required) Yes Manual Varies
Upstash (HTTP) Yes Yes Built-in 500K cmd/month

Key advantages:

  • HTTP-only transport — works in serverless functions and edge runtimes with no connection state
  • Auto-pipelining batches multiple commands into a single HTTP request automatically
  • Sliding window rate limiting built into the SDK (@upstash/ratelimit)
  • Full Redis command compatibility — not a subset or emulation
  • Free tier: 500K commands/month, 256MB storage

See vendors.md for pricing details and provider alternatives.

Known limitations

Deserialization surprises:

  • Numbers stored as strings come back typed as number, not string. Don't assume type roundtrip fidelity — validate on read.

Key scanning:

  • KEYS * scans the full keyspace synchronously. At 100K+ keys it blocks and degrades performance. Use SCAN with a cursor instead.

Missing hash keys:

  • hgetall returns null for a missing hash key, not an empty object {}. Guard against this before iterating.

Free tier inactivity:

  • Databases inactive for 14 days are archived and require manual restoration. Not suitable for infrequently accessed production data without a keep-alive strategy.

Consistency model:

  • Redis is not ACID. It is eventually consistent under replication. Do not use it as the source of truth for financial or transactional data.

Latency from region mismatch:

  • Upstash routes requests to the nearest replica. If your function region and Redis region differ, expect 500ms+ latency. Co-locate regions.

Related

  • postgres.md - Relational data, ACID transactions
  • neo4j.md - Graph data, relationship traversal
  • r2.md - Object storage for files and blobs
← Back to Stack