Joint control of compute and networking resources in future communication networks

Doktorandusz hallgató: 
Témavezető (TMIT): 
Doktori iskola: 
Informatikai Tudományok Doktori Iskola
Kiírás dátuma: 
2015. 01. 05
Téma leírása: 

Research objectives: Software-Defined Networking (SDN) and Network Functions Virtualization (NFV) are novel networking paradigms with several opportunities and challenges as well. By splitting the control and data planes, SDN makes much faster innovation possible in control plane applications, however, in a very different way than we followed so far. To understand this new approach and its impact on other components of the network is crucial. On the one hand, NFV aims at transforming traditional hardware based network functions and services into softwares which can be run on top of general purpose hardware elements such as commodity servers. On the other hand, cloud provides a virtualized computing infrastructure with the capability of running arbitrary software components. Combining cloud, NFV and SDN results in a network architecture which is able to jointly control and orchestrate compute and networking resources. Our main objectives are the following: design the main elements of this new architecture; understand the impact of virtualization solutions on the network itself and on the high level services; evaluate the performance of such virtualization techniques; propose novel methods and techniques to make control and data plane operation more effective.

Open problems/tasks:

  • Performance evaluation of compute control solutions; reveal drawbacks and performance issues of current methods; propose novel techniques, improvements; implement running prototypes.
  • Design and evaluate an architecture combining compute and networking resources into a common framework; propose orchestration algorithms jointly optimizing the usage of compute and networking resources; evaluate the algorithms according to defined metrics.
  • Analyze the performance of the virtualized data plane; reveal shortcomings of the approach; identify data plane functions which cannot be implemented efficiently in this framework; propose solutions that can mitigate the problem and able to improve performance.
Labor, csoport: 
Nagysebességű hálózatok laboratórium (HSN Lab)