1 / 7

Michael Welzl University of Oslo

The Benefits to Applications of using Explicit Congestion Notification (ECN) draft-welzl-ecn-benefits-00. Michael Welzl University of Oslo. Gorry Fairhurst University of Aberdeen. 89th IETF Meeting London, UK 4 March 2014. What are we doing?. Point of draft: document gains of ECN

harlan-paul
Download Presentation

Michael Welzl University of Oslo

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. The Benefits to Applications of using Explicit Congestion Notification (ECN)draft-welzl-ecn-benefits-00 Michael WelzlUniversity of Oslo Gorry FairhurstUniversity of Aberdeen 89th IETF Meeting London, UK4 March 2014

  2. What are we doing? • Point of draft: • document gains of ECN • includes less obvious gains • Out of scope: • To recommend a specific behavior

  3. ECN pro’s seem obvious • “The CE codepoint of an ECN-Capable packet SHOULD only be set if the router would otherwise have dropped the packet as an indication of congestion to the end nodes.” [RFC3168] • The receiver gets packets instead of loosing them. • Benefits shown in [RFC 2884], are not always biggest gain • In light to moderately loaded networks, number of dropped packets dropped due to congestion is small [RFC 3649] • …but that’s only a part of the story (Note: some but not all of ECN’s benefits need a different configuration than RFC 3168’s “SHOULD” above)

  4. Benefit from avoiding congestion loss • Reduced Head-of-Line Blocking • Relevant for transports providing in-order delivery • TCP/SCTP: Reduced Probability of RTO Expiry • RTO collapses cwnd, with significant bad impact; several mechanisms try to prevent this • Some applications do not retransmit lost packets • Typically VoIP, interactive video, real-time data • Need to apply loss-hiding mechanisms, with immediate effect on user-perceived quality

  5. Benefits that require special configuration • If ECN is configured such that routers mark packets at a lower level of congestion before they would otherwise drop packets from queue overflow: • Can avoid capacity overshoot; relevant e.g. in Slow Start • Can make congestion visible; relevant in ConEx • If a special configuration and reaction are used: • E.g. DCTCP has shown benefits when: • packets are marked earlier than they would otherwise be dropped • an instantaneous (= not averaged) queue is used for this decision (can be achieved with a special configuration of RED) • Receiver precisely feeds back number of ECN marks received in an RTT

  6. Conclusion • Motivates people configuring host stacks and network devices to enable ECN. • Application developers should where possible use transports that enable ECN. • Once enabled, the benefits of ECN are provided by the transport layer and the application does not need to be rewritten to gain these benefits.

  7. Next Steps • Is a document of this type helpful? • Are there other benefits to list? • Do we need to explain (potential) problems?

More Related