Why Pika Is Redefining Redis‑Compatible KV Storage on Disk

Pika is a disk‑based, fully Redis‑protocol‑compatible KV database that solves Redis memory bottlenecks. Developed by 360’s infrastructure and DBA teams, it was open‑sourced on GitHub around 2016 and later donated to the OpenAtom Open Source Foundation, where it was renamed OpenAtom Pika.

Thousands of developers from companies such as Sina, Kuaishou, Douyu, Ximalaya, Yuanfudao, and Huolala have adopted Pika, and a single deployment can handle billions of data items.

The project originated to reduce 360’s high data‑storage costs. As business grew, Redis demand increased, but memory‑based solutions caused long startup times, expensive master‑slave configurations, and costly hardware.

By storing data on disk while keeping a Redis‑compatible interface, Pika offers persistent storage, master‑slave replication, and both full and incremental synchronization.

Over five years, 360 operates more than 2,000 master‑slave instances and about ten Codis‑based Pika clusters, covering over 80 % of its services, including 360 Security Browser, 360 Search, 360 Cloud Disk, and Huajiao Live.

Industry observers note that Pika dramatically reduces server resource usage and improves data reliability, representing a from‑zero‑to‑one breakthrough for Redis‑like storage.

After joining OpenAtom, Pika seeks more contributors and plans to address industry trends, technical bottlenecks, and foundational breakthroughs to become a global KV‑database benchmark.

With the shift toward cloud‑native architectures, Pika—originally optimized for local SSD—must adapt to storage‑compute separation. Cloud disks provide larger capacity and lower cost but higher latency; future work includes optimizing Pika for cloud storage and exploring a write‑once‑multiple‑read architecture based on shared storage to further cut costs and enable elastic scaling.

The community launched the “Make Pika Great Again” initiative with partners such as Sina, Kuaishou, Douyu, Ximalaya, Yuanfudao, and Huolala to co‑plan development, optimization, and governance, aiming to build an international benchmark KV database.