Design
Distributed Metrics & Monitoring System
Design a monitoring system (like Prometheus) that collects, stores, and alerts on time-series metrics from thousands of services.
This is a Low-Level Design (LLD) problem. New to LLD vs HLD? Start here.
01 - Why interviewers ask this
It's the from-scratch "design Prometheus" question — interviewers check if you understand why time-series data needs specialized compression and downsampling, unlike generic row storage.
02 - Where this system exists in the real world
You interact with this design every day.
Prometheus / Datadog / Grafana monitoring stacks
Infrastructure observability platforms
Application performance monitoring (APM) tools
03 - What you'll master
Solve this once. Know it forever.
Understand time-series-specific compression
Design retention via downsampling tiers
Choose pull vs. push metric collection models
04 - What you'll design
6 functional · 3 non-functional requirements.
Collect numeric metrics (counters, gauges, histograms) from thousands of services
5 more requirements inside
Sign in to unlock the full brief and start designing
05 - Companies that ask this
You may face this exact question in your next interview.
Ready?
Open the canvas.
Design it yourself.
See all requirements, use staged hints, run your code, and compare with community discussions - all on the same canvas.
More in Infrastructure