Re: [aqm] Document Action: 'The Benefits of using Explicit Congestion Notification (ECN)' to Informational RFC (draft-ietf-aqm-ecn-benefits-08.txt)
Stuart, Sarma This draft describes the potential benefits when applications use a transport that enables Explicit Congestion Notification (ECN). Do you know if ECN provide savings in CPU cycles and memory usage , especially on the "server side" terminating a large number of TCP connections? Regards Steve B -Original Message- From: go...@erg.abdn.ac.uk [mailto:go...@erg.abdn.ac.uk] Sent: Wednesday, December 02, 2015 5:15 PM To: Steve Baillargeon Cc: Wesley Eddy; aqm-cha...@ietf.org; mls.i...@gmail.com; The IESG; draft-ietf-aqm-ecn-benef...@ietf.org; aqm@ietf.org Subject: RE: [aqm] Document Action: 'The Benefits of using Explicit Congestion Notification (ECN)' to Informational RFC (draft-ietf-aqm-ecn-benefits-08.txt) Does anyone have data? I'm not sure that it helps people to say there is research in evaluating the potential for ECN to save CPU cycles. I'm intrigued, and I'd like to see the research. But is this something we should add. Gorry > Hi Wes > I am happy with a note as a potential benefit for future exploration. > > -Steve > > -Original Message- > From: Wesley Eddy [mailto:w...@mti-systems.com] > Sent: Tuesday, December 01, 2015 10:03 PM > To: Steve Baillargeon > Cc: aqm-cha...@ietf.org; mls.i...@gmail.com; The IESG; > draft-ietf-aqm-ecn-benef...@ietf.org; aqm@ietf.org > Subject: Re: [aqm] Document Action: 'The Benefits of using Explicit > Congestion Notification (ECN)' to Informational RFC > (draft-ietf-aqm-ecn-benefits-08.txt) > > On 12/1/2015 5:22 PM, Steve Baillargeon wrote: >> Hi >> Sorry to come so late with a comment. >> Is it too late to add one more benefit to the draft? >> >> I suspect ECN brings potential and significant savings in CPU cycles >> and memory usage , especially on the "server side" terminating a >> large number of TCP connections. >> Has anyone done any analysis to confirm or contradict this assumption? >> > > > Hi Steve, thanks for the comment. > > I don't think I've seen anyone analyze that before, and would guess at > the moment that it's too tenuous to try to work into this particular > document at its advanced stage. > > I would recommend continuing discussion or research on this in AQM, > TSVWG, ICCRG or other appropriate groups at the moment, but not > altering the draft. At the ADs, and editors discretion, and if there > seems to be working group consensus, it might be noted as a potential > benefit for future exploration, but that's about the only impact I > think might be appropriate to this particular document at its advanced stage. > ___ aqm mailing list aqm@ietf.org https://www.ietf.org/mailman/listinfo/aqm
Re: [aqm] Document Action: 'The Benefits of using Explicit Congestion Notification (ECN)' to Informational RFC (draft-ietf-aqm-ecn-benefits-08.txt)
Hi Wes I am happy with a note as a potential benefit for future exploration. -Steve -Original Message- From: Wesley Eddy [mailto:w...@mti-systems.com] Sent: Tuesday, December 01, 2015 10:03 PM To: Steve Baillargeon Cc: aqm-cha...@ietf.org; mls.i...@gmail.com; The IESG; draft-ietf-aqm-ecn-benef...@ietf.org; aqm@ietf.org Subject: Re: [aqm] Document Action: 'The Benefits of using Explicit Congestion Notification (ECN)' to Informational RFC (draft-ietf-aqm-ecn-benefits-08.txt) On 12/1/2015 5:22 PM, Steve Baillargeon wrote: > Hi > Sorry to come so late with a comment. > Is it too late to add one more benefit to the draft? > > I suspect ECN brings potential and significant savings in CPU cycles and > memory usage , especially on the "server side" terminating a large number of > TCP connections. > Has anyone done any analysis to confirm or contradict this assumption? > Hi Steve, thanks for the comment. I don't think I've seen anyone analyze that before, and would guess at the moment that it's too tenuous to try to work into this particular document at its advanced stage. I would recommend continuing discussion or research on this in AQM, TSVWG, ICCRG or other appropriate groups at the moment, but not altering the draft. At the ADs, and editors discretion, and if there seems to be working group consensus, it might be noted as a potential benefit for future exploration, but that's about the only impact I think might be appropriate to this particular document at its advanced stage. ___ aqm mailing list aqm@ietf.org https://www.ietf.org/mailman/listinfo/aqm
Re: [aqm] Document Action: 'The Benefits of using Explicit Congestion Notification (ECN)' to Informational RFC (draft-ietf-aqm-ecn-benefits-08.txt)
Does anyone have data? I'm not sure that it helps people to say there is research in evaluating the potential for ECN to save CPU cycles. I'm intrigued, and I'd like to see the research. But is this something we should add. Gorry > Hi Wes > I am happy with a note as a potential benefit for future exploration. > > -Steve > > -Original Message- > From: Wesley Eddy [mailto:w...@mti-systems.com] > Sent: Tuesday, December 01, 2015 10:03 PM > To: Steve Baillargeon > Cc: aqm-cha...@ietf.org; mls.i...@gmail.com; The IESG; > draft-ietf-aqm-ecn-benef...@ietf.org; aqm@ietf.org > Subject: Re: [aqm] Document Action: 'The Benefits of using Explicit > Congestion Notification (ECN)' to Informational RFC > (draft-ietf-aqm-ecn-benefits-08.txt) > > On 12/1/2015 5:22 PM, Steve Baillargeon wrote: >> Hi >> Sorry to come so late with a comment. >> Is it too late to add one more benefit to the draft? >> >> I suspect ECN brings potential and significant savings in CPU cycles and >> memory usage , especially on the "server side" terminating a large >> number of TCP connections. >> Has anyone done any analysis to confirm or contradict this assumption? >> > > > Hi Steve, thanks for the comment. > > I don't think I've seen anyone analyze that before, and would guess at the > moment that it's too tenuous to try to work into this particular document > at its advanced stage. > > I would recommend continuing discussion or research on this in AQM, TSVWG, > ICCRG or other appropriate groups at the moment, but not altering the > draft. At the ADs, and editors discretion, and if there seems to be > working group consensus, it might be noted as a potential benefit for > future exploration, but that's about the only impact I think might be > appropriate to this particular document at its advanced stage. > ___ aqm mailing list aqm@ietf.org https://www.ietf.org/mailman/listinfo/aqm
[aqm] Document Action: 'The Benefits of using Explicit Congestion Notification (ECN)' to Informational RFC (draft-ietf-aqm-ecn-benefits-08.txt)
The IESG has approved the following document: - 'The Benefits of using Explicit Congestion Notification (ECN)' (draft-ietf-aqm-ecn-benefits-08.txt) as Informational RFC This document is the product of the Active Queue Management and Packet Scheduling Working Group. The IESG contact persons are Spencer Dawkins and Martin Stiemerling. A URL of this Internet Draft is: https://datatracker.ietf.org/doc/draft-ietf-aqm-ecn-benefits/ Technical Summary This document is mostly a list of demonstrated and expected benefits to transport protocols by using ECN. It highlights points that are most visible to the application layer within the end-points. It then goes on discussing specific deployment scenarios of ECN in a network, and the internet at scale. The key benefits of running ECN are summarized as o) Improved throughput o) Reduced Head-of-Line blocking o) Reduced probability of RTO Expiry o) Applications that do not retransmit lost packets o) Making incipient congestion visible o) Opportunities for new transport mechanisms Working Group Summary The document was brought to the working group to highlight and underline the many benefits ECN can have, if deployed at scale. During the WG discussions, the character of the draft changed slightly, from looking only at the positive implications to also describe potential drawbacks and pitfalls. The intention of this document though is less technical in nature, and instead is intended as a reference as to why deploying ECN at this time would be sensible. It aims to be a manifest that can be shown to decision-makers who quickly need to understand the key benefits of ECN, with a high level of technical guidance. Document Quality Are there existing implementations of the protocol? The document is agnostic of any specific implementation, and rather argues about the architectural model (well, as supported by the IP protocol) to use ECN. Implementations of ECN in TCP (RFC3168) are in wide-spread use, but with the ECN capabilities disabled, or only passively enabled. Arguably, the document helped to persuade decision-maker at a large vendor to actively start deploying ECN. Have a significant number of vendors indicated their plan to implement the specification? The document aims to achieve just that - to drive the adoption rate of a well known and available protocol by vendors up. Are there any reviewers that merit special mention as having done a thorough review, e.g., one that resulted in important changes or a conclusion that the document had no substantive issues? There were lively discussions in the AQM working group around this document. First, to not only speak exclusively about the positive aspects, but also mention potential issues. Second, that document had widespread support in the WG as it preaches to the choir - but word has to be spread about ECN to a larger audience. Personnel Who is the Document Shepherd? Richard Scheffenegger, AQM WG co-chair Who is the Responsible Area Director? Martin Stiemerling, Transport AD ___ aqm mailing list aqm@ietf.org https://www.ietf.org/mailman/listinfo/aqm
Re: [aqm] Document Action: 'The Benefits of using Explicit Congestion Notification (ECN)' to Informational RFC (draft-ietf-aqm-ecn-benefits-08.txt)
On 12/1/2015 5:22 PM, Steve Baillargeon wrote: Hi Sorry to come so late with a comment. Is it too late to add one more benefit to the draft? I suspect ECN brings potential and significant savings in CPU cycles and memory usage , especially on the "server side" terminating a large number of TCP connections. Has anyone done any analysis to confirm or contradict this assumption? Hi Steve, thanks for the comment. I don't think I've seen anyone analyze that before, and would guess at the moment that it's too tenuous to try to work into this particular document at its advanced stage. I would recommend continuing discussion or research on this in AQM, TSVWG, ICCRG or other appropriate groups at the moment, but not altering the draft. At the ADs, and editors discretion, and if there seems to be working group consensus, it might be noted as a potential benefit for future exploration, but that's about the only impact I think might be appropriate to this particular document at its advanced stage. ___ aqm mailing list aqm@ietf.org https://www.ietf.org/mailman/listinfo/aqm