- Version [version]
- Download 23
- File Size 626 KB
- File Count 1
- Create Date June 28, 2016
- Last Updated June 28, 2016
07. Header Compression in a Software Defined Network
Presented by William Flanagan, Flanagan Consulting.
Streaming media in the form of video dominates Internet traffic at times and is forecast to grow particularly quickly on mobile devices where bandwidth is limited on the Radio Access Network. Adding voice connections to packet networks is well underway and several organizations explicitly work toward dismantling the circuit-switched telephone network.
To date, packets on IP networks largely use version 4 addresses, but IPv6 inevitably will add many more bytes to each packet. For compressed voice, sending a sample of sound every 20 ms, the packet headers far exceed the payload yet carry little information that changes. The bandwidth consideration applies to video, though to a lesser extent as the payload of each packet can be larger. Software Defined Networks (SDNs) don't need the full stack of headers on every packet. The central controller can substitute very short flow IDs to guide payloads through a switched network, not only saving very significant bandwidth but also reducing latency at each switch. This presentation details the header compression ratios and estimates bandwidth reductions.