Why is Redis often faster than a traditional relational database for caching?

The speed difference comes from where and how data is stored and accessed, not from any single trick. Redis keeps its entire working dataset in RAM, so a read or write is a direct memory access rather than a disk seek, and its data structures (hash tables, skip lists, linked lists) are purpose-built for O(1) or O(log n) operations rather than the general-purpose relational model a SQL database has to support.

• In-memory storage — no disk I/O on the read/write path for normal operations, several orders of magnitude faster than disk access.
• Simple protocol — the RESP protocol is lightweight text/binary, avoiding the parsing and planning overhead of a SQL query.
• No query planner or joins — a GET or HGET is a direct structure lookup, not a query that needs to be parsed, planned, and optimized.
• Purpose-built data structures — a Sorted Set or Hash operation maps almost directly to an efficient underlying structure, rather than being expressed through general-purpose relational tables and indexes.

The trade-off is exactly what you'd expect from those design choices: Redis isn't meant to replace a relational database for complex, ad hoc queries, multi-table joins, or datasets far larger than available RAM — it excels specifically at fast, simple key-based access to a working set that fits in memory.