At a glance
- Did you know
- Providers can cache a long repeated prompt prefix, cutting cost and time to first token.
- Rule of thumb
- Put stable content first (instructions, documents) and the changing question last.
- Watch out
- Low hit rate on personalised prompts.
Its settings in ScaleMySystems
What you can change on it in the lab, and what each number does.
- Concurrency (slots)
- Calls it can work on at once (per instance when autoscaling). Needed slots ≈ calls per second × seconds each takes. Extra calls wait in the queue, then are rejected.
- Queue capacity
- Calls allowed to wait for a free slot. A bigger queue absorbs bursts but adds waiting time; 0 rejects straight away.
- vCPU
- CPU cores available. Needed ≈ calls per second × CPU per request ÷ 1000. Above about 90% busy it slows down and may crash.
- CPU per request (ms)
- CPU time each call burns. Part of what the component is; make the code cheaper to lower it.
- Latency (p50) (ms)
- Typical time one call takes here, before any calls it makes itself. Slots are held for this long.
- Latency (p99) (ms)
- The slowest 1% of calls. Callers must set their timeout above this or they cut off healthy calls.
- Timeout to dependencies (ms)
- How long it waits for each call it makes before giving up. Set it above the dependency’s p99 but low enough that a dead dependency does not hold every slot.
- Error rate
- Share of calls that fail on their own (bugs, declines, flaky tools). An agent that calls it on every step multiplies this.
- Quota (req/min)
- Calls allowed per minute (an API key or provider limit). Calls over it get 429s. Needed = calls per second × 60, with headroom.
- Hit rate
- Share of calls answered here without calling what sits behind it. Every miss is a call to the next component.
Problems that start with it
- LLM gateway with fallback 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.
- Model provider outage Your chat product depends on one model provider. At 20 s it is down for 15 seconds.
- Prompt cache that rarely hits An exact-match prompt cache sits in front of the model but hits only 5%: timestamps in the prompt make every prompt unique. The model’s token quota is tight.
Full lesson coming soon
Lessons are being added one component at a time: how it works, its settings, a live exercise to break and fix it, interview questions and a short check. See the Circuit breaker lesson for what one looks like.