CableLabs targets lag with Low-Latency DOCSIS

Sept. 16, 2019
Low-Latency DOCSIS (LLD) is on the move. CableLabs plans to host interoperability events this year and next year to iron out technology kinks and expects to see products ...

Low-Latency DOCSIS (LLD) is on the move. CableLabs plans to host interoperability events this year and next year to iron out technology kinks and expects to see products available in the next year.

The estimate is that with LLD technology, DOCSIS 3.1 and future equipment will provide latency of as low as 1 millisecond on the access network for those applications that require it. DOCSIS 3.1 as currently deployed offers latency of 10 milliseconds on the access network link. However, with a heavy load, this can jump to 100 milliseconds or more.

Queue-building applications, such as streaming video, cause latency because they send data as quickly as possible - faster than the link can support. The packets build up in the network.

"This is inherent in the design of the TCP protocol, which is used by many, many applications," said Greg White, distinguished technologist, CableLabs. "On the other hand, applications like gaming, that send one or two packets every 30th of a second, don't cause network latency, but their packets get caught behind these queues of TCP traffic."

With LLD, the user experience becomes more consistent because the non-queue-building apps can utilize a different path. A recent CableLabs [] blog post, co-authored by White, explained that LLD does not interfere with the sharing of total bandwidth, and it doesn't reduce one application's latency at the expense of others. Rather, those that send big-bursts of data can keep to the "classic" service, and those that demonstrate they do not queue-build can use the low-latency service.

The challenge, White said, is in properly identifying the non-queue-builders (NQB).

"Our view is that applications should identify their packets as NQB by tagging those packets with an identifier that is being standardized in the Internet Engineering Task Force (IETF). The network can then validate the marking and provide it with low latency treatment that will improve its performance. We recognize that broad-based application marking will be a challenge, so there are near-term solutions that can be implemented to perform marking on behalf of applications that don't support it yet."

CableLabs has been working with technology vendors in the CMTS, silicon and cable modem spaces to find a practical solution that can work with existing silicon, but may require new software in the CMTS and cable modems. The specs were issued in January 2019.

"We don't foresee any major challenges in implementing the technology, but we expect the usual minor issues around spec interpretation and different implementations that we will flush out during interops," White said. 

LLD also includes tools to enable automatic provisioning of the services and report latency performance statistics to the operator. Some low-latency features will also be available for pre-DOCSIS 3.1 cable modems.        

About the Author

BTR Staff

EDITORIAL
STEPHEN HARDY
Editorial Director and Associate Publisher
[email protected]
MATT VINCENT
Senior Editor
[email protected]
SALES
KRISTINE COLLINS
Business Solutions Manager
(312) 350-0452
[email protected]
JEAN LAUTER
Business Solutions Manager
(516) 695-3899
[email protected]

Sponsored Recommendations

How AI is driving new thinking in the optical industry

Sept. 30, 2024
Join us for an interactive roundtable webinar highlighting the results of an Endeavor Business Media survey to identify how optical technologies can support AI workflows by balancing...

Advances in Fiber & Cable

Oct. 3, 2024
November 7, 2024 1:00 PM ET / 12:00 PM CT / 10:00 AM PT / 6:00 PM GMT Duration: 1 hour Already registered? Click here to log in. A certificate of attendance...

Advancing Data Center Interconnection

July 25, 2024
Data Center Interconnect (DCI) solutions provide physical or virtual network connections between remote data center locations. Connecting geographically dispersed data centers...

The AI and ML Opportunity

Sept. 30, 2024
Join our AI and ML Opportunity webinar to explore how cutting-edge network infrastructure and innovative technologies can meet the soaring demands of AI memory and bandwidth, ...