What’s New in Valkey 9.0? Key Features and Performance Boosts Explained

Valkey, a fast‑growing Redis fork led by the Linux Foundation, released version 9.0 with a set of incremental yet impactful enhancements aimed at easing data management and operations.

Performance Optimizations

The release highlights support for Multi‑Path TCP (MPTCP) , allowing a single TCP connection to use multiple network paths simultaneously. This improves bandwidth aggregation and provides fault‑tolerance: if one path fails, traffic seamlessly switches to another, reducing replication latency and preventing downtime in high‑availability scenarios. Although official benchmarks are not yet published, the feature is expected to increase cluster resilience.

Additional low‑level tweaks include memory‑management and I/O optimizations that further boost throughput for demanding workloads.

New Commands

Hash Field Expiration

Traditional Redis hashes expire at the key level, which is coarse for use‑cases like session tokens versus user metadata. Valkey 9.0 adds HEXPIRE and HSETEX, enabling per‑field expiration without significant memory overhead.

HSETEX user:123 EX 900 FIELDS 1 auth_token "eyJhbGciOiJ..."

These commands are built on a lightweight “volatile set” structure, reducing stale data in micro‑service architectures.

DELIFEQ (Delete If Equal)

To simplify distributed lock handling, Valkey introduces DELIFEQ, which atomically deletes a key only if its value matches a given string, eliminating the need for Lua scripts.

DELIFEQ mylock ca304fec-bdd2-438d-8b69-3f9fa1ef9018

The command returns 1 on successful deletion and 0 otherwise, enhancing safety for patterns like Redlock.

SHUTDOWN SAFE

A new shutdown mode prevents a master node that still holds slot ownership from being stopped, protecting against accidental data loss during failover.

Cluster and Replication Enhancements

Valkey 9.0 improves the cluster model with atomic slot migration , ensuring slot transfers are indivisible and reducing consistency risks during scaling operations.

Multi‑database support is also refined, allowing smoother cross‑database operations within a cluster, which, combined with MPTCP‑enhanced replication, makes Valkey suitable for large‑scale e‑commerce and real‑time analytics.

Security and Ecosystem

While no brand‑new security patches are highlighted, features like DELIFEQ indirectly harden lock handling by removing script‑based vulnerabilities. Community contributions (e.g., GitHub PR #1671, #1949) underscore Valkey’s open‑source innovation beyond being a mere Redis clone.

The project’s growth illustrates how open‑source forks can drive rapid advancement in the in‑memory database ecosystem.

Conclusion

Valkey 9.0 delivers meaningful usability and operational reliability improvements, making it a compelling upgrade for high‑concurrency applications or existing Redis clusters. Future expectations include AI integration and edge‑computing support.