InterviewLegend

Design a Metrics Monitoring System

Signature problem
Alex Chen45 min readAdded 1,074 real interview reports

A metrics monitoring system collects the numbers that machines and services report about themselves, like CPU, memory, request counts and error rates. It stores them over time, draws them as graphs and tells someone when one goes wrong. Prometheus with Grafana, SignalFx and New Relic are products of this kind. LinkedIn asks something like this a lot, and the wording moves around. It might be "metric collection and querying", "metrics aggregation and visualization" or "monitor the health of services in real time". Underneath, each version asks for the same system. Points come in from a fleet, they get rolled up and stored, a dashboard reads them back, and a rule engine sends alerts. The write volume is what makes it hard. Almost every point is written once and never looked at by a human, unless something breaks.

The LinkedIn angle
LinkedIn has run its own version of this for over a decade: inGraphs for graphs, AutoAlerts for YAML rules, Kafka for transport. It shows up from IC2 to Staff, and one candidate was told to focus on the data store. Sketch the pipeline in ten minutes and keep most of the hour for storage, rollups and alerts.

LinkedInPremium lesson · System design round

The LinkedIn design, worked end to end.

  1. The API and data modelSystem design · Signature problem
  2. The design, drawn out2 diagrams
  3. 5 deep divesWith the tradeoffs
Get the LinkedIn questions

$49 a year · all 107 companies · instant access

1,074 LinkedIn interview reports · guide updated

LinkedIn interview guide

Know what LinkedIn asks before you walk in.

28 coding · 7 system design · 26 behavioral questions

Get Premium$49 a year · Instant access
Or start freeNo card.

Get their questions first.

Had your LinkedIn interview?How we research

InterviewLegend is independent and not affiliated with or endorsed by any company it covers.