home.social

#l4s — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #l4s, aggregated by home.social.

fetched live
  1. To learn how the sausage is made, see RFC9330 "Low Latency, Low Loss, and Scalable Throughput (L4S) Internet Service: Architecture"

    datatracker.ietf.org/doc/rfc93

  2. To learn how the sausage is made, see RFC9330 "Low Latency, Low Loss, and Scalable Throughput (L4S) Internet Service: Architecture"

    datatracker.ietf.org/doc/rfc93

    #networking #NetworkingProtocols #RFC #IETF #L4S

  3. The quiet plan to make the internet feel faster: engineers and major companies are pushing a technology called L4S that they say could make the web feel dramatically faster

    theverge.com/23655762/l4s-inte

    #web #internet #L4S #technology #IT

  4. The quiet plan to make the #internet feel faster --
    #Engineers and major companies are pushing a #technology called #L4S that they say could make the #web feel dramatically faster. But how?

    theverge.com/23655762/l4s-inte

  5. The quiet plan to make the #internet feel faster --
    #Engineers and major companies are pushing a #technology called #L4S that they say could make the #web feel dramatically faster. But how?

    theverge.com/23655762/l4s-inte

  6. The high-level summary is that #L4S has two main components:
    1) Communicating endpoints and network devices have a new way of using ECN. Rather than just indicating loss, proportional ECN can now indicate "how full" a link is.
    2) "Scalable" L4S flows and traditional flows use different AQM queues on network devices, with the former being very small.
    Then *LOTS* of details & corner cases, with a goal of L4S flows never having significant latency from queues on supporting devices/routers/etc.

  7. The high-level summary is that #L4S has two main components:
    1) Communicating endpoints and network devices have a new way of using ECN. Rather than just indicating loss, proportional ECN can now indicate "how full" a link is.
    2) "Scalable" L4S flows and traditional flows use different AQM queues on network devices, with the former being very small.
    Then *LOTS* of details & corner cases, with a goal of L4S flows never having significant latency from queues on supporting devices/routers/etc.

  8. L4S has the potential to be a big step towards reducing tail latency and improving responsiveness on the #Internet. The #IETF RFCs for #L4S have been published as #RFC 9330 (architecture), 9331 (experimental ECN changes for L4S), and 9332 (experimental DualQ AQM for L4S)! See rfc-editor.org/rfc/rfc9330.htm and friends. Low Latency DOCSIS builds on these already. Next steps are seeing how well these work (and/or break) when deployed at global scale. Thanks to hard work from Bob Briscoe and many others!

  9. L4S has the potential to be a big step towards reducing tail latency and improving responsiveness on the #Internet. The #IETF RFCs for #L4S have been published as #RFC 9330 (architecture), 9331 (experimental ECN changes for L4S), and 9332 (experimental DualQ AQM for L4S)! See rfc-editor.org/rfc/rfc9330.htm and friends. Low Latency DOCSIS builds on these already. Next steps are seeing how well these work (and/or break) when deployed at global scale. Thanks to hard work from Bob Briscoe and many others!