Next Generation Cloud-native In-Memory Stores: From Redis to Valkey and Beyond
Carl-Johan Fauvelle Munck af Rosensch"old, Feras M. Awaysheh, Ahmad Awad
TL;DR
This study provides an independent, end-to-end evaluation of Redis alternatives Valkey, KeyDB, and Garnet under realistic cloud-native workloads in Kubernetes. It combines quantitative performance benchmarking with practical migration and deployment analyses to illuminate trade-offs among throughput, latency, resource efficiency, compatibility, and long-term viability. Garnet delivers the highest raw performance at the cost of API compatibility, Valkey offers a compelling drop-in replacement with strong community backing, and KeyDB achieves moderate gains but raises sustainability questions. The work informs deployment decisions and highlights avenues for future research, including clustering, persistence, and cost-benefit analyses in production-like settings.
Abstract
In-memory key-value datastores have become indispensable building blocks of modern cloud-native infrastructures, yet their evolution faces scalability, compatibility, and sustainability constraints. The current literature lacks an experimental evaluation of state-of-the-art tools in the domain. This study addressed this timely gap by benchmarking Redis alternatives and systematically evaluating Valkey, KeyDB, and Garnet under realistic workloads within Kubernetes deployments. The results demonstrate clear trade-offs among the benchmarked data systems. Our study presents a comprehensive performance and viability assessment of the emerging in-memory key-value stores. Metrics include throughput, tail latency, CPU and memory efficiency, and migration complexity. We highlight trade-offs between performance, compatibility, and long-term viability, including project maturity, community support, and sustained development.