Dependability in a Multi-tenant Multi-framework Deep Learning as-a-Service Platform

TL;DR

This paper examines dependability challenges in a multi-tenant, multi-framework deep learning platform, proposing an architecture that balances reliability, scalability, and efficiency in cloud-based DL training environments.

Contribution

It introduces a dependability-focused architecture for a DLaaS platform, detailing design, implementation, and empirical evaluation of overheads and tradeoffs.

Findings

Dependability features improve fault tolerance in DL training.

The platform maintains scalability and flexibility with manageable overheads.

Tradeoffs between efficiency and dependability are identified and discussed.

Abstract

Deep learning (DL), a form of machine learning, is becoming increasingly popular in several application domains. As a result, cloud-based Deep Learning as a Service (DLaaS) platforms have become an essential infrastructure in many organizations. These systems accept, schedule, manage and execute DL training jobs at scale. This paper explores dependability in the context of a DLaaS platform used in IBM. We begin by explaining how DL training workloads are different, and what features ensure dependability in this context. We then describe the architecture, design and implementation of a cloud-based orchestration system for DL training. We show how this system has been architected with dependability in mind while also being horizontally scalable, elastic, flexible and efficient. We also present an initial empirical evaluation of the overheads introduced by our platform, and discuss tradeoffs between efficiency and dependability.