Redis 4.0 Unveiled: New Module System, PSYNC Enhancements, LFU Cache & More

Introduction

Redis 4.0 is about to be released, a major version with significant changes; here are several key new features.

Module System Introduced

The new module system allows custom extensions to Redis, enabling developers to create their own data types and functionalities.

For example, the Redis author has implemented a neural‑network data type, and others have built rate‑limiting commands, graph databases, secondary indexes, time‑series, full‑text search, and more.

With the module system, Redis can evolve beyond a simple database or cache; it could serve as a network server or a foundational protocol for various extensions.

Improved Replication Strategy

Previously, Redis used a fully synchronous replication: the master generated an RDB file, transferred it, and the slave loaded it.

Redis 4.0 introduces PSYNC with partial synchronization, where master and slave maintain replication offsets; after a disconnection, they can resume replication from the offset without a full sync.

PSYNC improves replication performance but has limitations during failover when a slave becomes master and other slaves lack offset records, requiring a full sync.

Version 4.0 adds a tag label combined with the offset, stored in the RDB file, so each slave records the tag and offset, enabling PSYNC‑based partial resynchronization even after a slave is promoted to master.

Cache Eviction Optimization

Memory is limited, and efficient cache reclamation is critical. Redis 4.0 refines the existing eviction strategy to be more robust, fast, and precise.

A new LFU (Least Frequently Used) policy is added to evict the least frequently accessed keys.

Non‑Blocking Deletion

The new UNLINK command performs a non‑blocking delete: it removes the key reference immediately and frees the memory asynchronously in a separate thread.

New MEMORY Commands

The MEMORY command suite provides detailed insight into memory usage, including:

Inspecting memory consumption of a specific key.

Viewing detailed memory usage breakdown.

Requesting memory release.

Examining internal allocator state.

Conclusion

The module system represents a major architectural shift, expanding Redis’s application scenarios, while the other enhancements boost performance and efficiency.

Beyond these highlights, Redis 4.0 includes additional features, making the upcoming release highly anticipated; the release candidate is already available, and the final version is imminent.