Planck: Millisecond-scale Monitoring and Control for Commodity Networks

Speaker:  Jeff Rasley, Brown University

Host:  Ivan Beschastnikh

Title:
Planck: Millisecond-scale Monitoring and Control for Commodity Networks

Abstract:
Software-defined networks introduce the possibility of building
self-tuning networks that constantly monitor network conditions and
react rapidly to important events such as congestion. Unfortunately,
state-of-the-art monitoring mechanisms require 100s of milliseconds to
seconds to extract global network state, like link utilization or the
identity of "elephant" flows.

This work introduces Planck, a network measurement architecture that
employs oversubscribed port mirroring to extract network information
at 275 µs – 4 ms timescales on a 10 Gbps network, 1–2 orders of
magnitude faster than recent approaches. We use this technique to
drive a traffic engineering application that can reroute congested
flows in milliseconds and obtain aggregate throughput within 1–4% of
optimal for most workloads we evaluated.

Bio:
Jeff Rasley is a 3rd year PhD student from Brown University advised by
Rodrigo Fonseca. He is supported by an NSF Graduate Research
Fellowship and interested in networks and distributed systems. He has
his undergraduate degree in Computer Science from the University of
Washington. http://cs.brown.edu/~jeffra/