: Latency reduction in the Internet - An Introduction to Bufferbloat mitigation and Flow-Aware Networking

Speaker:  Luca Muscariello – Paris, France
Topic(s):  Networks and Communications


Internet applications can have very different objectives when using network resources. For instance, bulk transfer of data compared to real-time communications have very different objectives as they tend to optimize throughput and latency respectively. These different goals, sometimes in conflict one with the other, have significant impact on design choices of transport protocols and traffic management mechanisms.
For many complex reasons, that we will briefly review in this presentation, the IPv4 Internet has turned into a communication network that can carry almost only TCP traffic. However, if we observe that current applications are predominantly Cloud-native, highly distributed and composed of an heterogeneous traffic mix of real-time and bulk transfer applications, the problem remains open. Solution like diffserv can help in very specific deployments only which do not include Cloud-native applications, wireless and mobile networks.
In this talk we give an introduction of some of the most powerful techniques to optimize different applications objectives, in the transport layer and also by using the most recent traffic management techniques based on flow-aware networking, per-flow packet scheduling and active queue management.  We will show that these techniques can be very effective in an "ossified" IPv4Internet and provide a very powerful tool for the IPv6 Internet. Considering a technical literature that spans over four decades, the talk will focus on most recent applications and network deployments with major emphasis to what is currently available in the Internet and ready to use.

About this Lecture

Number of Slides:  50
Duration:  60 minutes
Languages Available:  English
Last Updated: 

Request this Lecture

To request this particular lecture, please complete this online form.

Request a Tour

To request a tour with this speaker, please complete this online form.

All requests will be sent to ACM headquarters for review.