hardDesignScenarioMidSeniorStaff
Five months after a machine failure, an engineer asks you for the high-rate data around it. Was your retention policy right?
7 min readUpdated 2026-08-15Target archetype: Enterprise Captive, Product Startup
Answer
Short answer
Retention is a bet about which questions get asked later, and averaging is the bet that loses, because the transient that explains a failure disappears into a mean. Keep min, max and count per bucket rather than an average, and let events freeze a window of full-rate data at the edge.
Keep historian explicit in the answer because that is the concept the interviewer is actually trying to test. A good historian explanation names the trade-off, the failure mode, and the evidence you would use before choosing.
Was this answer helpful?
Likely follow-ups
- Your collector uses a deadband. What class of fault does that make invisible, and how would you find out?
- How would you retain a vibration waveform without retaining every waveform?
- The engineer wants the data and also wants to know it was not altered. What does that add?
- Two years of raw data exists and nobody can query it usefully. What have you actually got?
Related questions
- Design the ingestion path for sensor telemetry from a few hundred machines. Which decisions matter most?hardAlso on downsampling and retention6 min
- How do you handle personal data in prompts, logs and traces for an LLM feature?mediumAlso on retention5 min
- How do you design a system so an auditor can prove who accessed a record and why?mediumAlso on retention8 min
- A legal hold lands on Friday afternoon and your nightly deletion jobs run at two. What do you do, and what do you build afterwards?hardAlso on retention6 min