Resilient Communication with Programmable Hardware

• Applicant: Professor Dr. Michael Menth
• Subject Area: Security and Dependability, Operating-, Communication- and Distributed Systems ; Communication Technology and Networks, High-Frequency Technology and Photonic Systems, Signal Processing and Machine Learning for Information Technology
• Term: since 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 503231190

Resilience features in communication protocols mitigate critical situations where, e.g., availability, security, or performance are at risk. They are indispensable for deployment but increase complexity. In this work we focus on the design of communication protocols with resilience features and their implementation on devices with programmable data planes (forwarding hardware). Devices with programmable data plane support the implementation of existing and new protocols on commodity hardware and are key ingredients for disaggregation of communication infrastructure. On the one hand, their operations per packet are limited on high-speed platforms so that efficient implementations are challenging and feasible only for protocols with moderate complexity. On the other hand, they can assist the control and data plane through new packet headers, flexible packet processing, and optimizations. The latter use software-defined networking with learning algorithms that automatically adapt configurations to current network conditions. Successful innovations on devices with programmable data plane that are suitable for a larger market can be implemented on specialized ASICs in the future. For this work, we choose resilience mechanisms on high-speed platforms as use cases that are relevant for today’s network operation and that are discussed in standardization bodies. Those are resilient stateless multicast, 1+1 protection with packet loss recovery, hardware-assisted message flooding under unreliable forwarding conditions, mitigation of denial-of-service attacks against mapping-based communication, and automated deployment of crypto-protocols. The proposed resilience features benefit from the advantages of programmable hardware and software-defined control, but require novel system architectures and protocols, optimized operation through online learning, scalability analyses, and in particular running prototypes to demonstrate their feasibility. The project is involved in IETF standardization to disseminate findings, get feedback on early results, and shape the evolution of future protocols. The work explores limits of today’s devices with programmable data plane and proposes hardware extensions to support attractive use cases. Finally, efficient blueprints of resilient protocols on programmable hardware are provided which are useful for disaggregation.

• DFG Programme: Priority Programmes
• Subproject of: SPP 2378: Resilience in Connected Worlds – Mastering Failures, Overload, Attacks, and the Unexpected