DeepRecSys: A System for Optimizing End-To-End At-scale Neural Recommendation Inference

TL;DR

DeepRecSys introduces a co-designed system and scheduler that significantly improves the efficiency and latency of neural recommendation inference at scale, benefiting cloud infrastructure and services.

Contribution

The paper presents DeepRecInfra and DeepRecSched, novel end-to-end system and scheduling solutions tailored for recommendation inference, achieving doubled throughput and reduced latency.

Findings

System throughput doubled across eight models

Over 30% latency reduction in production datacenter

Effective handling of diverse inference query patterns

Abstract

Neural personalized recommendation is the corner-stone of a wide collection of cloud services and products, constituting significant compute demand of the cloud infrastructure. Thus, improving the execution efficiency of neural recommendation directly translates into infrastructure capacity saving. In this paper, we devise a novel end-to-end modeling infrastructure, DeepRecInfra, that adopts an algorithm and system co-design methodology to custom-design systems for recommendation use cases. Leveraging the insights from the recommendation characterization, a new dynamic scheduler, DeepRecSched, is proposed to maximize latency-bounded throughput by taking into account characteristics of inference query size and arrival patterns, recommendation model architectures, and underlying hardware systems. By doing so, system throughput is doubled across the eight industry-representative recommendation models. Finally, design, deployment, and evaluation in at-scale production datacenter shows over 30% latency reduction across a wide variety of recommendation models running on hundreds of machines.