At a glance
- Did you know
- An LLM judge grades outputs at scale, but it is another model call with its own latency, cost and biases.
- Rule of thumb
- Judge a sample asynchronously instead of every answer inline.
- Watch out
- Doubles model calls; slow judges back up the pipeline.
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.
Problems that start with it
- An LLM judge on every answer Every answer is graded inline by an LLM judge with 20 slots before it returns. The judge is now the slowest part.
- Corrective RAG evaluator bottleneck Corrective RAG grades retrieved chunks with an evaluator model before answering. The evaluator has 20 slots and grades every question.
- Content moderation queue Every post is checked by a moderation model with 30 slots, about 600 ms per post, at 80 posts a second.
- Judging every agent step To evaluate agents, an LLM judge scores every step of every trajectory (was the tool call right, did the step help) while tasks run. 8 tasks a second × 6 steps reach a judge with 20 slots at about 0.6 s.
- ChatGPT: from launch build to real traffic A chat product on launch-week hardware: the GPU prefill and decode pools, the KV cache store, moderation and the chat API are an eighth of the size real traffic needs. 300 chats a second arrive with long prompts. Answers queue for 20 seconds and most fail.
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.