Title: Disaggregated Optical Networks: hype, huge potential or somewhere in the middle?

Organizers: Ramon Casellas (CTTC), Josep Maria Fàbrega (CTTC), Filippo Cugini (CNIT)

Description: Optical disaggregation roughly involves utilizing whitebox components without compromising overall network performance. Through disaggregation, network operators can have flexible and fine-grained control over their network from control and management to the network node hardware. Upgrading any network nodes or network software with cutting-edge technologies, customizing them or even adding specialized hardware or software is easier and more cost efficient. Additionally, disaggregation allows independent evolution of different components of the network with instant deployment.

As of today, the benefits of optical disaggregation have not been fully established, and some concerns are raised regarding disaggregation. On the one hand, it is important to clearly identify the shortcomings and define in which contexts it does not make sense to opt for disaggregation. On the other hand, there are challenges open to fully exploit disaggregation benefits. For example, the concept of network disaggregation has been tied to the use of AI/ML techniques for aspects related to optical transmission and network operations. However, protocols and frameworks initial applicability mainly targeted device configuration and simple monitoring and notification support, and AI/ML require availability of high volumes of data that need to be collected and processed. Aspects such as speed of data retrieval and processing, latency and bandwidth constraints or data synchronization requirements are becoming increasingly relevant in large-scale telemetry and ML driven network operation.

This workshop aims at covering the role of disaggregation for a multiple perspective

  • Can current SDN controllers and frameworks address the complexity of some disaggregation models such as full disaggregation? Do the systems scale with a large number of devices? Is the current monolithic format of SDN controllers a bottleneck, and should we consider new deployments based e.g. on cloud native solutions and containers?
  • Telemetry for disaggregated networks still presents important challenges, such as limitations to the currently adopted modeling languages, an increasing number of overlapping yet incompatible models or the lack of consensus regarding the optimal set of telemetry data that needs to be made available and exported. What advances can be made in monitoring and telemetry systems that are not expensive and can be deployed at a massive scale, yet working at high data rates? Can telemetry data be made open while still allowing for vendor optimization? How can operators use such high volume of data efficiently? What protocols and interfaces seem the most adequate for telemetry and (real-time) monitoring given bandwidth and latency constraints? What advances can be made in monitoring and telemetry systems that are not expensive and can be deployed at a massive scale, yet working at high data rates? Can telemetry data be made open while still allowing for vendor optimization? How can operators use such high volume of data efficiently? What protocols and interfaces seem the most adequate for telemetry and (real-time) monitoring given bandwidth and latency constraints?
  • Techno-economical analysis showing the benefits of network disaggregation is still at early stages. What are the actual benefits? Are proprietary vendors subsidizing the cost of Open Line Systems aiming at obtaining benefits through selling transceivers?
  • Can emerging new applications (e.g. Optical Ethernet, Deterministic Networking) rely on disaggregated networks?

 

 

We use cookies to facilitate navigation and improve your experience across our website. By clicking "Accept", you will be storing these cookies.