Grampín Castro, Eduardo

Research interests

Internet routing
Datacenter networking Control Plane
network programmability

Working area

My activity includes teaching, research, extension and university management, within the framework of Cyber-Physical Systems (CPS), which link the physical world and computing and communications, with feedback loops between both worlds. My area of ​​research is Network and Computing Infrastructure for CPS. Given the evolution of the Internet towards a network architecture based on content provider datacenters, since my research and teaching stay at Roma Tre University in 2019, I have started to explore new research topics linked to the so-called "Masive Scale Data Centers” (MSDC), in particular interconnection topologies based on non-blocking Clos networks, and new routing mechanisms, which combine features of traditional "Link State" and "Distance Vector" protocols. In this framework, my activity has focused on analyzing the applicability and management problems of traditional protocols (such as BGP) and new proposals, including monitoring, and data-driven decision making and AI/ML (Machine Learning) techniques, which have become ubiquitous at all levels. The aim is to make network equipment generic and programmable, and to actively participate in these processes, giving rise to "Hardware-accelerated Machine Learning", which takes advantage of their programmability. Indeed, specific programming languages ​​such as P4 are being developed that allow functionalities to be implemented according to application needs at line speed on network processors (nanoseconds instead of the milliseconds it takes the CPU). In this context, SmartNICs and programmable switches are becoming increasingly common. Developing network management and control solutions in this context poses significant challenges that we are investigating together with undergraduate and graduate students of whom I am a supervisor.