hard20 pts · first answer countsmodel-servingautoscaling
Autoscaling an LLM deployment
GPU pods running an LLM server are autoscaled on CPU usage, which stays at 15% even while users wait 20 seconds. Which signal is better for scaling?
Discussion 0
Join free to post in the discussion.
Related questions
- When batch inference fitsMLOps Engineer
- Traffic before the model has loadedMLOps Engineer
- Why export a model to ONNXMLOps Engineer
- Protocol for internal inference callsMLOps Engineer
- Latency and throughputMLOps Engineer
- What TTFT measuresMLOps Engineer
- Gap between streamed tokensMLOps Engineer
- Two phases of LLM generationMLOps Engineer