Special CESG Seminar – Maple: Simplifying SDN Programming Using Algorithmic Policies

Special CESG Seminar – Maple: Simplifying SDN…

Special CESG Seminar – Maple: Simplifying SDN…
CESG Seminar: Catapulting beyond Moore’s Law: Using FPGAs to Accelerate Data Centers

CESG Seminar: Catapulting beyond Moore’s Law:…

CESG Seminar: Catapulting beyond Moore’s Law:…
CESG Seminar: Hardware Implementation of Cascade Support Vector Machine

CESG Seminar: Hardware Implementation of Casc…

CESG Seminar: Hardware Implementation of Casc…
CESG Fishbowl Tele-Seminar: What are They doing with Your Data?

CESG Fishbowl Tele-Seminar: What are They doi…

CESG Fishbowl Tele-Seminar: What are They doi…
ILP and TLP in Shared Memory Applications: a Limit Study Dissertation by: Ehsan Fatehi

ILP and TLP in Shared Memory Applications: a …

ILP and TLP in Shared Memory Applications: a …
PreviousNext

Special CESG Seminar – Maple: Simplifying SDN Programming Using Algorithmic Policies

Last week, Dr. Andreas Voellmy gave a talk about software defined networking (SDN). Dr. Voellmy gave an overview of SDN and presented several challenges with OpenFlow, a standard for SDN. He also presented Maple, which addresses these challenges

A recent development in networking, Software Defined Networking allows a network to make changes to its behavior through a central policy administered by a network controller. While previously, network architecture consisted of fixed, closed, vertically-integrated network appliances, SDN implements a more general packet processing approach, programmed through open control software executed on servers. This implementation is fairly open and very flexible. One standard for SDN is OpenFlow, which defines certain rules and guidelines for how SDN should be implemented. Many aspects of this implementation remain challenging, however.

To address these challenges, Dr. Voellmy presented Maple. Maple allows the user to create algorithmic policies, algorithms programmed in some general-purpose language and run on every packet of data that enters a network. These algorithmic policies replace the requirement of SDN to generate and maintain sets of rules on individual network switches. To implement these policies, Maple has a tracing runtime system which discovers reusable forwarding decisions from a control program.

Dr. Andreas Voellmy received his PhD in Computer Science from Yale University. His research focuses on Software Defined Networking, where he mainly draws on the OpenFlow library implementation, and the Glasgow Haskell Compiler.