Clipper: A Low-Latency Online Prediction Serving System
This addresses the need for efficient and robust online prediction serving in applications with heavy query loads, representing an incremental improvement over existing serving systems.
The paper tackles the problem of deploying machine learning models for real-time prediction serving by introducing Clipper, a low-latency system that simplifies deployment across frameworks and applications. It demonstrates that Clipper reduces latency and improves throughput, accuracy, and robustness, achieving comparable performance to TensorFlow Serving while enabling model composition and online learning.
Machine learning is being deployed in a growing number of applications which demand real-time, accurate, and robust predictions under heavy query load. However, most machine learning frameworks and systems only address model training and not deployment. In this paper, we introduce Clipper, a general-purpose low-latency prediction serving system. Interposing between end-user applications and a wide range of machine learning frameworks, Clipper introduces a modular architecture to simplify model deployment across frameworks and applications. Furthermore, by introducing caching, batching, and adaptive model selection techniques, Clipper reduces prediction latency and improves prediction throughput, accuracy, and robustness without modifying the underlying machine learning frameworks. We evaluate Clipper on four common machine learning benchmark datasets and demonstrate its ability to meet the latency, accuracy, and throughput demands of online serving applications. Finally, we compare Clipper to the TensorFlow Serving system and demonstrate that we are able to achieve comparable throughput and latency while enabling model composition and online learning to improve accuracy and render more robust predictions.