Part 1 of 8 · 1 min read
Why this page comes second in Level 12
In this module you will
- Scope an internal metrics system and size its write rate and storage with roll-ups.
- Explain a time series, a metric name and labels, and why a general database is a poor fit.
- Compare pull and push collection, and use service discovery and consistent hashing for pull.
- Place a queue and aggregation in the pipeline, and choose where to aggregate.
- Describe the alert flow, with rules, a manager that merges alerts, and notification channels.
Level 9 explained what to measure and when to alert. This page builds the system that does it. It shows a pipeline with heavy constant writes and bursty reads, which is the opposite of most web products. It also uses a queue (page 1 of this level) to protect the database. The book used is "System Design Interview, Volume 2" by Alex Xu, the metrics monitoring and alerting chapter. Numbers marked "made up" only show the idea.