Holger Karl

Holger Karl is currently full professor of “Internet Technology and Softwarization” at the Hasso Plattner Institute, University of Potsdam. Until summer of 2021, he held the chair of Computer Networks at Paderborn University. His research interests lie in making networks more flexible and using them as a computing platform, in the sense of edge clouds, as well as in wireless and mobile communication. Recently, he has looked at applying machine learning as a means to optimize network operations as well as at the challenges ensuing from deploying machine learning pipelines inside a network.

Keynote: “Network Softwarization – Some orchestration problems”
Abstract: In the past, networks were simply intended to forward packets from source to destination. Later, networks started to perform more complex actions on packets or data flows, e.g., stateful firewalls or video re-encoding. Typically, these functions were implemented in hardware, with typical pros (efficient, fast) and cons (slow time to market, difficult to change). Softwarization refers to instead running such network-internal functions in software, on top of a (more or less) general-purpose execution infrastructure. But once such an infrastructure is in place, it can be used to run general-purpose services as well – this opened the door to the notion of edge clouds, in-network clouds, etc. In such a context, a couple of challenges arise, e.g., where to run which component of such a service – these questions are summarized as “orchestration”. This talk will look at a couple of orchestration problems and discuss various challenges, e.g., scalability.