The Internet is based on fundamental design goals that emphasize simplicity and reliability. Early on, the idea was that instead of a centrally controlled network that was a gatekeeper on new features (such as telephony ‘star’ codes) most of the intelligence was decentralized into the edge systems to help enable a tremendous wave of permissionless innovation at the application layer. Any complex network-oriented processing was to be performed by the host devices. A fundamental difference of today’s Internet is ‘predictable service qualities’, which can be achieved only by ‘smart’ components in an end-to-end path. Over the years, applications have adapted as well. A classic success story is Internet streaming. Through maintaining an appropriate size playback buffer and aided with adaptive bitrate control, HTTP-based adaptive streaming (and UDP equivalents) is the dominant application.
As society moves from the information age to the age of the M2M communications, we anticipate further significant changes to the Internet as well as significantly different applications. Machines are making decisions based on data, many times in real-time. There are wired and wireless scenarios to consider. Many of these applications have yet to be invented, but based on early examples such as cloud gaming, enhanced videoconferencing, coordinated autonomous vehicles, drone swarms working in a coordinated manner to carry out missions, multiuser virtual-reality gaming, the Internet needs to be re-examined to determine what must be addressed so that technological breakthroughs in devices and systems are not held-back by Internet performance limitations. These concerns have motivated the key standards bodies to pursue broad initiatives that will enhance their respective technology’s ability to support emerging applications systems that require predictable service qualities.
Cable operators are evaluating Low Latency DOCSIS (LLD) to determine how the new technology might improve subscriber’s perceived quality and how an operator might leverage the technology for new services [1, 2, 3].
In this paper, we first present LLD lab analysis that shows promising results. We then discuss end-to-end factors that are critical in providing predictable end-to-end LL services. We conclude our paper with an architecture including network and service components to deploy an effective LL system.