Problems / Async and streaming, hard
Asked at Datadog, Google and Netflix. Every metric point is inserted into one column store, and dashboards query the same raw points. At 10 s an incident doubles the points (and everyone opens dashboards).
Unlock it once for 25 credits; everything inside it is free (runs, changes, Submit, results, hints). Sign in with Google to start; new accounts get free credits.
Success ≥ 99.9% and p99 ≤ 1 s, for at most $32,000 a month.
Like a coding problem, a solution has to pass every case, not just the one it was built for. Submit in the lab runs all 3; 1 of them are hidden until then.
40,000 requests a second for 20 seconds: 98% ingest (writes), 2% query (reads).
What each component does in this design, and what calls it. The simulator mimics each one from these settings.
| Component | Its job here | Settings |
|---|---|---|
| Load balancerLoad balancer | Traffic enters here.Calls Ingest API, Query API.Layer 4 balancers route by IP and port; layer 7 balancers read HTTP and can route by path or header. | – |
| Ingest APIApp service | Load balancer forwards to it for /ingest.Calls Raw points (one column store).Little's law: throughput ≈ concurrency ÷ latency. 200 slots at 50 ms serve about 4,000 req/s. | 16 replicas × 100 slots · timeout 1.0 s |
| Query APIApp service | Load balancer forwards to it for /query.Calls Raw points (one column store).Little's law: throughput ≈ concurrency ÷ latency. 200 slots at 50 ms serve about 4,000 req/s. | 2 replicas × 100 slots · timeout 4.0 s |
| Raw points (one column store)Columnar analytics DB | Ingest API calls it. Query API calls it.Columnar storage compresses well because each column holds similar values. | 256 slots |