Using deep programmability to put network owners in control

Abstract

Controlling an opaque system by reading some "dials" and setting some "knobs," without really knowing what they do, is a hazardous and fruitless endeavor, particularly at scale. What we need are transparent networks, that start at the top with a high-level intent and map all the way down, through the control plane to the data plane. If we can specify the behavior we want in software, then we can check that the system behaves as we expect. This is impossible if the implementation is opaque. We therefore need to use open-source software or write it ourselves (or both), and have mechanisms for checking actual behavior against the specified intent. With fine-grain checking (e.g., every packet, every state variable), we can build networks that are more reliable, secure, and performant. In the limit, we can build networks that run autonomously under verifiable, closed-loop control. We believe this vision, while ambitious, is finally within our reach, due to deep programmability across the stack, both vertically (control and data plane) and horizontally (end to end). It will emerge naturally in some networks, as network owners take control of their software and engage in open-source efforts; whereas in enterprise networks it may take longer. In 5G access networks, there is a pressing need for our community to engage, so these networks, too, can operate autonomously under verifiable, closed-loop control.

Deep data plane programming and AI for zero-trust self-driven networking in beyond 5G

“…Software-defined, programmable networking fabric is recognized as a key enabling technology to support 5G and beyond networks in achieving their promises of high-level scalability and flexibility with lower CAPEX and OPEX [21]. Going beyond 5G will be also characterized by pursuing deep programmability of the network fabric both vertically (i.e., control and data plane) and horizontally (i.e., end-toend from the radio access network to edge and core network) [53] to fulfill the stringent performance requirements. In fact, endowing the next-generation mobile networks with data plane programming capabilities will bring various benefits, including dynamic traffic engineering at wire-speed, INT for latency-critical services, slicing and multi-tenancy, in-network security, and offloading of 5G VNF to data plane [54].…”

A Survey on Machine Learning Techniques for Routing Optimization in SDN

“…Now that most network innovation in companies is based on open source software, we, as part of the research community, should profit from it and leverage the same platforms and tools for a more effective adoption by industry. Alternatively, merging efforts with other big projects like Pronto [199], [200] would be clearly beneficial. Additionally, considering the application of ML in routing is usually foreseen as a step towards automatized network management, we should continuously monitor to what extent is ML trusted by network operators.…”

An Exhaustive Survey on P4 Programmable Data Plane Switches: Taxonomy, Applications, Challenges, and Future Trends

“…Future work should explore more sophisticated methods for solving the partitioning problem, while considering the constraints of the hardware and the current networking landscape. Note that tremendous efforts from academia and the industry are being spent on the Pronto project [343,344], which can be considered as an example of the deep programmable architecture.…”