Understanding and Solving Hot Key Issues in Redis

1 What is a Hot Key?

In Redis, a hot key refers to a key that receives a very high request frequency, e.g., millions of requests in a short time, which can concentrate traffic and potentially cause the Redis server to crash.

Subsequent requests then fall back to the backend database, whose performance may be insufficient, risking overload and service unavailability.

2 Causes of Hot Keys

1. Consumer demand far exceeds production, such as flash sales, trending news, or popular comments (read‑heavy, write‑light scenarios).

During events like Double‑11 flash sales, a single product may be accessed or purchased hundreds of thousands of times, generating massive Redis traffic and creating hot‑key problems.

2. Request sharding concentrates on a single Redis node, exceeding its performance limit.

When data is partitioned by a fixed hash, many keys may map to the same Redis instance; a sudden traffic spike can overwhelm that node, causing a hot‑key issue.

3 Risks of Hot Keys

Cache breakdown, Redis server overload, and a flood of requests to the backend service, which may trigger a cascade failure (snowball effect) if the database cannot handle the load.

4 How to Identify Hot Keys

1. Use domain knowledge and business scenarios to guess potential hot keys.

For example, during a flash‑sale of iPhone models, the SKU for the iPhone becomes a hot key.

2. Instrument client code to count key accesses and report statistics.

Modify business code to record each key’s usage before accessing Redis, aggregate the data periodically, and identify hot keys; this approach adds some intrusion to the code.

3. Collect metrics at the service proxy layer.

If a proxy sits in front of Redis, it can gather usage data and help locate hot keys; architecture varies across companies.

4. Use Redis built‑in commands.

Commands such as --hotkeys with redis-cli or the MONITOR command can capture real‑time traffic, though they may impact performance and should be used cautiously.

5. Packet capture and analysis on Redis nodes.

Write custom programs to listen on the Redis port, parse traffic, and analyze hot‑key patterns.

5 Solutions to Hot Key Problems

1. Expand the Redis cluster: add more shards and replicas to distribute read traffic.

2. Implement a secondary cache (e.g., JVM local cache) to reduce Redis reads.

For instance, using Caffeine together with Redis forms a two‑level cache: first check the local cache, then fall back to Redis; alternatives include Ehcache or even a simple HashMap.