Technical Conference: 07 - 11 March 2027
Exhibition: 09 - 11 March 2027
Los Angeles Convention Center | Los Angeles, California, United States

Technical Conference: 07 - 11 March 2027
Exhibition: 09 - 11 March 2027
Los Angeles Convention Center | Los Angeles, California, United States

Data Center Interconnects for Hyper-scale Cloud & AI Networking

SC461 - Data Center Interconnects for Hyper-scale Cloud & AI Networking

15 Mar 2026
08:30 - 12:30

Short Course Level

Beginner

Short Course Description

AI and cloud networking have driven demand for data center interconnects (DCI) to unprecedented levels in recent years. At the same time, technology development on optics and electronics has allowed for the integration of state-of-the-art optical transceivers in small, pluggable, form factors for direct use in routing and switching equipment.

This short course gives a broad overview of DCI architectures and technology in hyper-scale AI and cloud networking, ranging from short-haul interconnects of just a few kilometers across a campus, to metro and finally long-haul deployments. We focus, in particular, on DC edge architectures for front-end networks to connect DCs within a metropolitan region. In addition, we will focus on “scale across” architectures for back-end networks, which are used to interconnect AI clusters. 

We explain in this short course how the latest advances in DWDM transport enable the realization of ever denser, more power efficient, higher capacity and more scalable data center interconnect links. We will provide an overview of the different technologies that are most relevant for today’s data center interconnects, covering the full range of transmitter and receiver technologies, pluggable form factors, modulation formats, protocols, and data rates. We will also cover the latest industry developments, with a special focus on ZR/ZR+ at 400G and 800G, and future 1.6T, pluggable DWDM optics, and how this impacts edge DCI architectures.

Besides transceiver technology, we will also cover system design aspects such as data center architectures, open line systems, disaggregation of the optical layer and cloud-scale network management and monitoring. We will discuss distinctives of data center interconnects when compared to traditional optical transport networks, and what are the most relevant design criteria in today’s DCI deployments. Finally, we will cover in detail the latest industry standardization initiatives around pluggable DWDM optics.

The short course should enable attendees to obtain a detailed overview of the different technologies and architectures that are most relevant to data center interconnects and get a comprehensive understanding for the different system design trade-offs in terms of cost, capacity, density, power consumption and complexity.

Short Course Benefits

This course should enable you to:
 

  • Understand what drives the need for data center interconnects. How does DCI play a role in modern data center architectures, and what is the impact of new use cases such as distributed AI cluster training and inference.
  • Describe the typical requirements for data center interconnects, including transmission distances, system capacity, traffic profiles, data rates, protocols and transmission latency.
  • Understand the system architecture of today’s data center interconnects, and which architectures are typically utilized in high-capacity data center interconnect deployments at 400G and 800G, as well as how this will evolve at 1.6T data rates and beyond.
  • Be able to understand industry trends toward IP-optical integration with ZR/ZR+ pluggable DWDM optics, and how this can be used to reduce power and space requirements for short-reach DCI interconnects.
  • Describe how network disaggregation, open line systems, network monitoring, and management are implemented in cloud-scale networks.
  • Understand the different transceiver architectures in the context of short-reach DCI, for example direct detection in comparison to coherent detection, and how to leverage the trade-off between optical performance and system complexity / cost.
  • Obtain an overview of the different standardization bodies relevant for data center interconnects such as IEEE, OIF, OpenZR+ and OpenROADM MSA, and how they contribute to the next generation of DC architectures at 1.6T and beyond.

Short Course Audience

This course is intended for engineers, researchers and technical managers who would like to gain a better understanding of data center interconnect technology, architectures and solutions as used in hyperscale cloud & AI networking. Practical network implementations and system design trade-offs will be discussed in detail, such that the participants can obtain a good level of understanding for the challenges of today’s data center interconnect solutions. Participants should have a comprehensive knowledge in the field of fiber-optic transmission systems; no previous knowledge of data center architectures is required.

Instructors

  • Mark Filer

    Oracle, United States

    About the Instructor

    Mark M. Filer works at Oracle as architect and area tech lead in optics and photonic interconnect for AI and public cloud systems. Immediately prior, Mark was at a stealth startup where he served as area tech lead in photonic architectures for GW-scale AI systems, focusing on driving innovations and ecosystems to support power-/cost-optimized and operationally resilient AI networks. He has spent much of his career at hyperscalers (Google, Microsoft and AWS) where he worked across diverse areas of optical networking including DC-fabric and AI backend systems, metro/DCI, long-haul and subsea. Mark has consistently participated in industry standard and consortia organizations to specify and deliver open platforms (such as 400ZR), and he had the honor of serving on the OIF Board of Directors as Vice President. He has served on OFC technical committees, Optica Executive Forum committee, and as a reviewer for IEEE and Optica journal publications. Mark has authored several publications and patents in the areas of photonic integration, cloud-scale network architectures, open line systems, long-haul transmission, and DWDM system impairments. He holds B.Sc. and M.Sc. degrees in electrical engineering from the Georgia Institute of Technology in Atlanta, GA.

  • Dirk van den Borne

    Hewlett Packard Enterprise, Germany

    About the Instructor

    Dirk van den Borne is a director of system engineering at Hewlett Packard Enterprise, where he leads a “Networking for AI” team of solution architects that advise operators on technology evolution across data center switching, WAN routing and transport. He specializes himself in the convergence of IP and optical transport, and how developments in Ethernet, optical integration, and telemetry/analytics are shaping network architectures.
    Dirk obtained his Ph.D. in optical communications from the Eindhoven University of Technology. He has spoken frequently at major industry events, authored and co-authored over 100 peer-reviewed journal and conference contributions and holds several patents on optical communication. He served on various technical program committees, including 7 years on the OFC committee. He is based in Munich, Germany.