Disaggregated Accelerator Management System for Cloud Data Centers

TL;DR

This paper introduces Flow-in-Cloud, a disaggregated data center architecture that enhances resource flexibility and utilization by dynamically managing compute and accelerator resources over a high-speed network.

Contribution

It proposes a novel disaggregated architecture and management system for accelerators in cloud data centers, demonstrated through a proof of concept with deep learning applications.

Findings

Feasibility of disaggregated accelerator pools demonstrated

Successful deployment of distributed deep learning on prototype system

Enhanced resource management flexibility shown

Abstract

A conventional data center that consists of monolithic-servers is confronted with limitations including lack of operational flexibility, low resource utilization, low maintainability, etc. Resource disaggregation is a promising solution to address the above issues. We propose a concept of disaggregated cloud data center architecture called Flow-in-Cloud (FiC) that enables an existing cluster computer system to expand an accelerator pool through a high-speed network. FlowOS-RM manages the entire pool resources, and deploys a user job on a dynamically constructed slice according to a user request. This slice consists of compute nodes and accelerators where each accelerator is attached to the corresponding compute node. This paper demonstrates the feasibility of FiC in a proof of concept experiment running a distributed deep learning application on the prototype system. The result successfully warrants the applicability of the proposed system.