1 / 7

Advancing RFC 4138 < draft-ietf-tcpm-rfc4138bis-01 > < draft-kojo-tcpm-frto-eval-01 >

Advancing RFC 4138 < draft-ietf-tcpm-rfc4138bis-01 > < draft-kojo-tcpm-frto-eval-01 >. Pasi Sarolahti Markku Kojo Kazunori Yamamoto Max Hata IETF-70 / TCPM / Vancouver, BC, Canada / December 4 th 2007. Problems with regular TCP. On Spurious Timeouts

malina
Download Presentation

Advancing RFC 4138 < draft-ietf-tcpm-rfc4138bis-01 > < draft-kojo-tcpm-frto-eval-01 >

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. 1 Advancing RFC 4138<draft-ietf-tcpm-rfc4138bis-01><draft-kojo-tcpm-frto-eval-01> Pasi Sarolahti Markku Kojo Kazunori Yamamoto Max Hata IETF-70 / TCPM / Vancouver, BC, Canada / December 4th 2007

  2. 2 Problems with regular TCP • On Spurious Timeouts • Regular TCP sender retransmits whole window unnecessarily in slow start • Network resources are wasted • In many cases severe performance penalty to the TCP flow • Dishonors packet conservation principle

  3. 3 F-RTO: Detecting Spurious RTOs • F-RTO slightly modifies TCP sender behavior • After RTO retransmission try to send a couple of new segments • If new acknowledgements for non-retransmitted segments flow in, assume RTO was spurious • Otherwise new segments trigger DupACKs, and sender should assume genuine RTO • No TCP options required • Compatible with existing TCP implementations • Does not cause network congestion • Might not detect spurious timeout in some cases • If F-RTO does not detect spurious RTO, it reverts back to traditional RTO recovery

  4. 4 Current Progress • Revised RFC 4138 targeting at PS <draft-ietf-tcpm-rfc4138bis-01> • Changes from -00: • Added back the original SACK-algorithm from RFC 4138 after the common feedback to have the SACK-algorithm in the document • Clarified behavior on multiple timeouts • Clarified that ACKs that do not acknowledge new data but are not duplicate acknowledgements are ignored • Other small clarifications on both algorithms and general editing • Added one paragraph describing the basic idea of the SACK algorithm

  5. 5 Current Progress (cont’d) • Wrote I-D"Evaluation of RFC 4138" <draft-kojo-tcpm-frto-eval-01.txt> • Points out the problems with regular RTO recovery and usefulness of F-RTO • Evaluates F-RTO to show it is not harmful to the network, corner cases included • Summarizes experimentation results • Changes from -eval-00: • Added a summary on experimentation with malicious receiver • receiver does not benefit from cheating when conservative response is used • receiver may benefit when aggressive response is used • General editing

  6. 6 Ready to advance? • A number of known F-RTO implementations are out there • Proposals and support to advance to PS have been expressed several times by implementors • Experimentations have been carried with several implementations showing positive results • All feedback has been positive • Implementors: straightforward to implement no issues with the spec • Many implementations enable F-RTO by default • Windows Vista • Microsoft report at IETF-68 about their positive experiences • Linux • SACK-enhanced F-RTO enabled by default from up-coming release of 2.6.24 and onward, and falls back to basic variant if SACK not negotiated

  7. 7 Next Steps • Basically ready for WGLC • First need green light from WG for advancing to PS

More Related