[Paper Review] Mission Apollo: Landing Optical Circuit Switching at Datacenter Scale

Ryohei Urata, Hong Liu+12 more|arXiv (Cornell University)|Aug 22, 2022

Advanced Photonic Communication Systems24 citations

TL;DR

The paper presents Apollo, the world’s first large-scale production deployment of optical circuit switches (OCSes) for datacenter networking, detailing the hardware design, bidirectional linking with circulators, and co-design with WDM transceivers to meet bandwidth, scale, and cost targets.

ABSTRACT

In this paper, we describe Apollo, to the best of our knowledge, the world's first large-scale production deployment of optical circuit switches (OCSes) for datacenter networking. We will first describe the infrastructure challenges and use cases that motivated optical switching inside datacenters. We then delve into the requirements of OCSes for datacenter applications: balancing cost, port count, switching time, and optical performance, which drive design choices and implementation details of our internally developed 3D MEMS-based OCS. To enable the Apollo optical switching layer, we employ circulators to realize bidirectional links through the OCS, effectively doubling the OCS radix. The OCS and circulator design choices were critical for meeting network bandwidth, scale, and cost targets. We review the critical co-design of WDM transceiver technology for these OCS plus circulator-based bidirectional links and their corresponding physical impairments, delivered over four generations/speeds of optical interconnect. Finally, we conclude with thoughts on future directions in hardware development and associated applications.

Motivation & Objective

• Motivate the use of optical circuit switching inside datacenters by outlining infrastructure challenges and use cases.
• Identify the key requirements for OCS deployments in terms of cost, port count, switching time, and optical performance.
• Describe the design and implementation decisions of the 3D MEMS-based OCS and supporting components to meet target metrics.

Proposed method

• Develop a 3D MEMS-based optical circuit switch (OCS) suitable for datacenter scale.
• Use circulators to realize bidirectional links through the OCS to effectively double the OCS radix.
• Co-design WDM transceivers with the OCS and circulator-based links to address physical impairments across multiple generations/speeds of optical interconnect.
• Analyze hardware design choices and their impact on bandwidth, scale, and cost targets.

Experimental results

Research questions

• RQ1 How can optical circuit switching be effectively deployed at datacenter scale in production environments?
• RQ2 What are the key design tradeoffs (cost, port count, switching time, optical performance) for OCS-based datacenter networks?
• RQ3 How do circulators enable bidirectional links and how does this affect OCS radix and overall network performance?
• RQ4 What are the implications of co-designing WDM transceivers with OCS and circulator-based links across multiple generations?

Key findings

• Apollo represents the world’s first large-scale production deployment of OCSes in datacenters.
• Circulators enabling bidirectional links through the OCS are critical for achieving higher effective radix and network reach.
• Co-design of WDM transceivers with OCS and circulator-based links is essential to address physical impairments across generations of optical interconnects.
• Hardware design choices were shaped to meet network bandwidth, scale, and cost targets across four generations/speeds of optical interconnects.