Problems / AI systems, medium
40 chats/s through your AI gateway and a prompt cache to one hosted model. At 20 s the provider has a 15 s outage.
Free to open. Runs in your browser, no sign-up.
Keep success ≥ 99% and time to first token p99 ≤ 2.5 s through the outage. Add a fallback to a second provider.
40 requests a second for 50 seconds: 100% chat (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 AI gateway (keys, routing, retries).Layer 4 balancers route by IP and port; layer 7 balancers read HTTP and can route by path or header. | – |
| AI gateway (keys, routing, retries)App service | Load balancer forwards to it.Calls Prompt cache.Little's law: throughput ≈ concurrency ÷ latency. 200 slots at 50 ms serve about 4,000 req/s. | 2 replicas × 400 slots · timeout 20 s |
| Prompt cachePrompt cache | AI gateway (keys, routing, retries) calls it.Calls Primary model (provider A).Providers can cache a long repeated prompt prefix, cutting cost and time to first token. | 2000 slots · 20% hit rate |
| Primary model (provider A)LLM endpoint | Prompt cache calls it.Latency has two parts: time to the first token, then a steady stream of tokens per second. | 400 streams · 3,000 calls/min |