Session organized by IEEE Future Directions
This panel will review the main bandwidth drivers of the cloud network evolution. In many respects, the explosive growth of “cloud” service delivery has been the biggest evolution in networking since the proliferation of the Internet. The corresponding growth of Data Center Interconnect (DCI) has arguably been the most significant evolution in wireline transport since the transitions from TDM to IP/MPLS and WDM. Motivated by the initial cloud services, mostly related to virtual compute and storage, the traffic interconnecting data centers has grown more than any other network traffic type, and has already dominated the global fiber infrastructure. The economics of this growth has, in turn, motivated the building of dedicated global DCI networks, that scale to some of the most spectrally dense fiber deployments. It also motivated a new class of purpose‐built routing and optical transport systems which have pioneered the adoption of routers focusing on maximizing throughput (rather than routing scale), WDM systems with state‐of‐the‐art coherent transmission, as well as software/SDN innovations in automation, programmability, management abstraction, and control‐plane disaggregation, and the associated “open” transport architectures.
So, what will drive the next cloud network evolution? New cloud services, including cloud‐based high‐performance computing, machine‐learning and AI, or the onset of quantum computing, often coupled with new applications like analytics, IoT, and augmented, virtual, and digital reality, have been suggested as significant new drivers for the growth of cloud networks, which are projected to grow by at least two orders of magnitude in the next few years. This evolution has also motivated discussions about innovations in the cloud network architectures, and the related systems and technologies. This panel will debate (beyond hype) the bandwidth requirements that are expected to drive the evolution and innovations in cloud networks. Some of the key questions the panel will debate include:
- What is the current reality and realistic potential of the emerging cloud services like Machine Learning and AI?
- Will Machine to Machine traffic continue to grow exponentially compared to Machine to
- End User? What are the similarities and differences in cloud network requirements from these new services and their related applications when compared to the current cloud infrastructure ‐ and what may be the role of optical transport in the cloud evolution?
- Are there important technology limitations or key innovations that would impact/enable the cloud evolution? For example, are there any potential synergies with the current IEEE Future Directions initiatives and activities in areas such as Digital Reality, Quantum Computing, Rebooting Computing, Cybersecurity, Fog/Edge, Big Data, Roadmaps and Blockchain?
- Are standards helping or hindering the evolution of cloud networks in response to the bandwidth drivers?