A multi-tenant analytics platform ingests events from 3,000 customer accounts
A medium Pipeline Design mock interview question on DataDriven. Practice with AI-powered feedback, real code execution, and a hire/no-hire decision.
- Domain
- Pipeline Design
- Difficulty
- medium
Interview Prompt
A multi-tenant analytics platform ingests events from 3,000 customer accounts. One enterprise tenant occasionally bursts to 30 percent of total throughput and slows ingest latency for the other 90 percent of tenants. There is no per-tenant rate cap and no per-tenant lag dashboard. The section names three isolation layers: logical (tenant id stamped on every record, shared infra), resource (each tier gets its own queue or worker pool), and physical (largest tenants get dedicated infrastructure). Plus fair scheduling (round-robin or weighted-fair across tenant-keyed sub-queues). Isolate the noisy tenant by adding the right isolation layer for this platform's tier structure (free, pro, enterprise) and a per-tenant lag monitor that surfaces who is impacting whom.
How This Interview Works
- Read the vague prompt (just like a real interview)
- Ask clarifying questions to the AI interviewer
- Write your pipeline design solution with real code execution
- Get instant feedback and a hire/no-hire decision