LY Corporation, the Japanese technology conglomerate formed from the merger of Yahoo! Japan and Korean messaging service LINE, is undertaking a significant overhaul of its cloud infrastructure. The company announced plans to replace its heavily customized OpenStack cloud with a more conventional version, consolidating its resources into a single cloud environment named “Flava.”
Consolidation Strategy
The transition to Flava is driven by the need for a unified cloud capable of supporting LY’s extensive services, which include the LINE messaging app and Yahoo portal, catering to approximately 300 million monthly users. Previously, LINE operated an internal cloud called “Verda,” which utilized 130,000 virtual machines (VMs) across 11,000 hosts within four OpenStack clusters. Meanwhile, Yahoo! Japan’s cloud, known as “YNW,” managed over 160,000 VMs across more than 160 OpenStack clusters on 27,000 servers.
New Cloud Architecture
LY Corporation’s new cloud architecture will streamline operations by reducing the number of hosts to 500 or more and limiting the VMs to over 9,000, all under a single OpenStack cluster. Ryuutarou Inoue, head of LY’s Cloud Infrastructure Unit, stated that the previous cloud’s extensive custom modifications made upgrades challenging. The new approach minimizes custom patches and encourages contributions to the upstream OpenStack project, facilitating regular updates and enhancing security.
Operational Improvements
Inoue outlined three core pillars of the Flava design: pursuing statelessness by treating VM root disk data as temporary; ensuring application-driven availability by integrating infrastructure with application architecture; and enabling faster recovery through Infrastructure as Code (IaC) practices. This strategy emphasizes maintaining service continuity rather than restoring previous states during incidents.
Focus on Observability and Automation
LY Corporation is also prioritizing observability in its cloud operations. The team employs tools like Prometheus and Grafana to monitor cloud health and detect anomalies. Inoue noted that hardware failures occur daily, necessitating automation in failure detection and recovery processes. The company aims to further enhance automation by incorporating large language models into decision-heavy workflows.
This consolidation comes in the wake of significant information security issues that have previously exposed user data, prompting government intervention to improve LY’s technology stack for better security and privacy.
This article was produced by NeonPulse.today using human and AI-assisted editorial processes, based on publicly available information. Content may be edited for clarity and style.
