5 Essential Redis Tips from Heroku to Boost Performance

Here are five practical recommendations from Heroku for using Redis effectively.

1. Use a connection pool

Using a connection pool reduces performance loss from excessive connections and lowers the number of connection creations, improving performance. Most Redis client libraries already provide connection pool implementations, such as redis3m (C++), gore (Go), jedis (Java), redis-connection-pool (Node.js), etc. Just enable the pool and set an appropriate pool size.

2. Set a name for the client

The CLIENT SETNAME command lets you assign a name to a client, making the output of CLIENT LIST more readable. This helps you quickly identify which client executed which commands and monitor memory usage when troubleshooting performance issues.

3. Choose an appropriate key eviction policy

By default Redis does not evict keys, which can cause errors when memory is insufficient. Redis offers several eviction policies, such as allkeys-lru (evicts any key using LRU) and volatile-lru (evicts only keys with an expiration time using LRU). See the maxmemory-policy setting in the configuration file for more options.

4. Avoid using the KEYS command

While useful for development, KEYS performs an O(N) scan and can degrade performance proportionally to the number of results. In production, consider using the SCAN command instead.

5. Set a suitable connection timeout

Redis does not close idle connections by default ( timeout 0). It is recommended to configure a non‑zero timeout value appropriate for your application to free idle connections.