The Missing Piece in Complex Analytics: Low Latency, Scalable Model Management and Serving with Velox

TL;DR

Velox is a scalable, low-latency system designed to manage, maintain, and serve complex statistical models in real-time, bridging the gap between offline training and online deployment in large-scale analytics.

Contribution

This paper introduces Velox, a novel system that enables efficient online model management and serving, addressing a critical gap in existing data analytics pipelines.

Findings

Velox achieves low-latency model serving at scale.

It supports lightweight online model maintenance and dynamic weighting.

Velox effectively integrates online and offline systems for real-time analytics.

Abstract

To support complex data-intensive applications such as personalized recommendations, targeted advertising, and intelligent services, the data management community has focused heavily on the design of systems to support training complex models on large datasets. Unfortunately, the design of these systems largely ignores a critical component of the overall analytics process: the deployment and serving of models at scale. In this work, we present Velox, a new component of the Berkeley Data Analytics Stack. Velox is a data management system for facilitating the next steps in real-world, large-scale analytics pipelines: online model management, maintenance, and serving. Velox provides end-user applications and services with a low-latency, intuitive interface to models, transforming the raw statistical models currently trained using existing offline large-scale compute frameworks into full-blown, end-to-end data products capable of recommending products, targeting advertisements, and personalizing web content. To provide up-to-date results for these complex models, Velox also facilitates lightweight online model maintenance and selection (i.e., dynamic weighting). In this paper, we describe the challenges and architectural considerations required to achieve this functionality, including the abilities to span online and offline systems, to adaptively adjust model materialization strategies, and to exploit inherent statistical properties such as model error tolerance, all while operating at "Big Data" scale.