Using SLI/SLO to Improve Reliability - Part 1: Building an SLI/SLO Framework and Developing LINE Status
Key point
LINE presents how it systematically manages organizational service reliability using an SLI/SLO framework and LINE Status.
Details
LINE's SRE team found a common pattern while applying SLI/SLO across multiple services. Most services went through the same stages from SLI/SLO adoption to operation.
Based on this, LINE built a 5-stage framework:
- Define: Selecting Critical User Journeys (CUJ) and defining SLIs
- Instrument: Designing and implementing metrics to measure SLIs
- Observe: Setting up Grafana dashboards and recording rules
- React: Setting SLO target values and configuring alerts (recommended: 99.9% on a 28-day rolling window)
- Improve: Error budget-based operations and regular reviews
It is currently provided as a Confluence template, with each team defining their SLI/SLO by completing each stage.
Based on this framework, an internal tool called LINE Status was developed. Unlike existing external API status pages, it automatically updates internal service status according to consistent criteria (SLI/SLO metrics).
The core design philosophy is to display "whether SLOs are met based on SLIs, instead of binary status (up/down)." It also avoids directly exposing technical metrics, instead expressing things in terms of functionality that users perceive. For a messaging service, terms like "message sending" and "read receipts" are used.
Key screens in LINE Status:
- Main page: Shows the current status of all services at a glance. AI-generated summaries, status per CUJ item, intuitive color codes (green = normal, yellow = event detected, red = outage)
- Service detail page: Status per CUJ item and event timeline. Monthly event history
- History page: All service events. Ability to grasp the scope of impact per event
A notable point during development was that "when collaborating with AI, it's important to clearly write requirements in Markdown." The author, as an engineer, didn't often do frontend work, but by documenting requirements in detail for each feature and sharing them with AI in a "vibe coding" approach, completed the UI in about 1 month.
Through the SLI/SLO framework and LINE Status, LINE aims to establish a common language for reliability discussions across the organization. Instead of individual teams' monitoring dashboards, the goal is to jointly assess user experience impact through CUJ-based SLI/SLO, enabling faster decision-making and communication between teams.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.