Efficient Serving of LLM Applications with Probabilistic Demand Modeling
This addresses the challenge of resource management for LLM application serving systems, offering a domain-specific improvement over existing blackbox approaches.
The paper tackles the problem of inefficient serving of LLM applications due to dynamic resource demands by proposing Hermes, a system that uses a Probabilistic Demand Graph model to optimize scheduling and backend prewarming, resulting in over 70% reduction in average completion time and over 80% reduction in P95 completion time.
Applications based on Large Language Models (LLMs) contains a series of tasks to address real-world problems with boosted capability, which have dynamic demand volumes on diverse backends. Existing serving systems treat the resource demands of LLM applications as a blackbox, compromising end-to-end efficiency due to improper queuing order and backend warm up latency. We find that the resource demands of LLM applications can be modeled in a general and accurate manner with Probabilistic Demand Graph (PDGraph). We then propose Hermes, which leverages PDGraph for efficient serving of LLM applications. Confronting probabilistic demand description, Hermes applies the Gittins policy to determine the scheduling order that can minimize the average application completion time. It also uses the PDGraph model to help prewarm cold backends at proper moments. Experiments with diverse LLM applications confirm that Hermes can effectively improve the application serving efficiency, reducing the average completion time by over 70% and the P95 completion time by over 80%.