One document matched: draft-ietf-dccp-tfrc-faster-restart-05.ps
%!PS-Adobe-3.0
%%BoundingBox: 24 24 588 768
%%Title: Enscript Output
%%For: Sally Floyd
%%Creator: GNU enscript 1.6.1
%%CreationDate: Sun Nov 18 21:49:46 2007
%%Orientation: Portrait
%%Pages: (atend)
%%DocumentMedia: Letter 612 792 0 () ()
%%DocumentNeededResources: (atend)
%%EndComments
%%BeginProlog
%%BeginResource: procset Enscript-Prolog 1.6 1
%
% Procedures.
%
/_S { % save current state
/_s save def
} def
/_R { % restore from saved state
_s restore
} def
/S { % showpage protecting gstate
gsave
showpage
grestore
} bind def
/MF { % fontname newfontname -> - make a new encoded font
/newfontname exch def
/fontname exch def
/fontdict fontname findfont def
/newfont fontdict maxlength dict def
fontdict {
exch
dup /FID eq {
% skip FID pair
pop pop
} {
% copy to the new font dictionary
exch newfont 3 1 roll put
} ifelse
} forall
newfont /FontName newfontname put
% insert only valid encoding vectors
encoding_vector length 256 eq {
newfont /Encoding encoding_vector put
} if
newfontname newfont definefont pop
} def
/SF { % fontname width height -> - set a new font
/height exch def
/width exch def
findfont
[width 0 0 height 0 0] makefont setfont
} def
/SUF { % fontname width height -> - set a new user font
/height exch def
/width exch def
/F-gs-user-font MF
/F-gs-user-font width height SF
} def
/M {moveto} bind def
/s {show} bind def
/Box { % x y w h -> - define box path
/d_h exch def /d_w exch def /d_y exch def /d_x exch def
d_x d_y moveto
d_w 0 rlineto
0 d_h rlineto
d_w neg 0 rlineto
closepath
} def
/bgs { % x y height blskip gray str -> - show string with bg color
/str exch def
/gray exch def
/blskip exch def
/height exch def
/y exch def
/x exch def
gsave
x y blskip sub str stringwidth pop height Box
gray setgray
fill
grestore
x y M str s
} def
% Highlight bars.
/highlight_bars { % nlines lineheight output_y_margin gray -> -
gsave
setgray
/ymarg exch def
/lineheight exch def
/nlines exch def
% This 2 is just a magic number to sync highlight lines to text.
0 d_header_y ymarg sub 2 sub translate
/cw d_output_w cols div def
/nrows d_output_h ymarg 2 mul sub lineheight div cvi def
% for each column
0 1 cols 1 sub {
cw mul /xp exch def
% for each rows
0 1 nrows 1 sub {
/rn exch def
rn lineheight mul neg /yp exch def
rn nlines idiv 2 mod 0 eq {
% Draw highlight bar. 4 is just a magic indentation.
xp 4 add yp cw 8 sub lineheight neg Box fill
} if
} for
} for
grestore
} def
% Line highlight bar.
/line_highlight { % x y width height gray -> -
gsave
/gray exch def
Box gray setgray fill
grestore
} def
% Column separator lines.
/column_lines {
gsave
.1 setlinewidth
0 d_footer_h translate
/cw d_output_w cols div def
1 1 cols 1 sub {
cw mul 0 moveto
0 d_output_h rlineto stroke
} for
grestore
} def
% Column borders.
/column_borders {
gsave
.1 setlinewidth
0 d_footer_h moveto
0 d_output_h rlineto
d_output_w 0 rlineto
0 d_output_h neg rlineto
closepath stroke
grestore
} def
% Do the actual underlay drawing
/draw_underlay {
ul_style 0 eq {
ul_str true charpath stroke
} {
ul_str show
} ifelse
} def
% Underlay
/underlay { % - -> -
gsave
0 d_page_h translate
d_page_h neg d_page_w atan rotate
ul_gray setgray
ul_font setfont
/dw d_page_h dup mul d_page_w dup mul add sqrt def
ul_str stringwidth pop dw exch sub 2 div ul_h_ptsize -2 div moveto
draw_underlay
grestore
} def
/user_underlay { % - -> -
gsave
ul_x ul_y translate
ul_angle rotate
ul_gray setgray
ul_font setfont
0 0 ul_h_ptsize 2 div sub moveto
draw_underlay
grestore
} def
% Page prefeed
/page_prefeed { % bool -> -
statusdict /prefeed known {
statusdict exch /prefeed exch put
} {
pop
} ifelse
} def
% Wrapped line markers
/wrapped_line_mark { % x y charwith charheight type -> -
/type exch def
/h exch def
/w exch def
/y exch def
/x exch def
type 2 eq {
% Black boxes (like TeX does)
gsave
0 setlinewidth
x w 4 div add y M
0 h rlineto w 2 div 0 rlineto 0 h neg rlineto
closepath fill
grestore
} {
type 3 eq {
% Small arrows
gsave
.2 setlinewidth
x w 2 div add y h 2 div add M
w 4 div 0 rlineto
x w 4 div add y lineto stroke
x w 4 div add w 8 div add y h 4 div add M
x w 4 div add y lineto
w 4 div h 8 div rlineto stroke
grestore
} {
% do nothing
} ifelse
} ifelse
} def
% EPSF import.
/BeginEPSF {
/b4_Inc_state save def % Save state for cleanup
/dict_count countdictstack def % Count objects on dict stack
/op_count count 1 sub def % Count objects on operand stack
userdict begin
/showpage { } def
0 setgray 0 setlinecap
1 setlinewidth 0 setlinejoin
10 setmiterlimit [ ] 0 setdash newpath
/languagelevel where {
pop languagelevel
1 ne {
false setstrokeadjust false setoverprint
} if
} if
} bind def
/EndEPSF {
count op_count sub { pos } repeat % Clean up stacks
countdictstack dict_count sub { end } repeat
b4_Inc_state restore
} bind def
% Check PostScript language level.
/languagelevel where {
pop /gs_languagelevel languagelevel def
} {
/gs_languagelevel 1 def
} ifelse
%%EndResource
%%BeginResource: procset Enscript-Encoding-88591 1.6 1
/encoding_vector [
/.notdef /.notdef /.notdef /.notdef
/.notdef /.notdef /.notdef /.notdef
/.notdef /.notdef /.notdef /.notdef
/.notdef /.notdef /.notdef /.notdef
/.notdef /.notdef /.notdef /.notdef
/.notdef /.notdef /.notdef /.notdef
/.notdef /.notdef /.notdef /.notdef
/.notdef /.notdef /.notdef /.notdef
/space /exclam /quotedbl /numbersign
/dollar /percent /ampersand /quoteright
/parenleft /parenright /asterisk /plus
/comma /hyphen /period /slash
/zero /one /two /three
/four /five /six /seven
/eight /nine /colon /semicolon
/less /equal /greater /question
/at /A /B /C
/D /E /F /G
/H /I /J /K
/L /M /N /O
/P /Q /R /S
/T /U /V /W
/X /Y /Z /bracketleft
/backslash /bracketright /asciicircum /underscore
/quoteleft /a /b /c
/d /e /f /g
/h /i /j /k
/l /m /n /o
/p /q /r /s
/t /u /v /w
/x /y /z /braceleft
/bar /braceright /tilde /.notdef
/.notdef /.notdef /.notdef /.notdef
/.notdef /.notdef /.notdef /.notdef
/.notdef /.notdef /.notdef /.notdef
/.notdef /.notdef /.notdef /.notdef
/.notdef /.notdef /.notdef /.notdef
/.notdef /.notdef /.notdef /.notdef
/.notdef /.notdef /.notdef /.notdef
/.notdef /.notdef /.notdef /.notdef
/space /exclamdown /cent /sterling
/currency /yen /brokenbar /section
/dieresis /copyright /ordfeminine /guillemotleft
/logicalnot /hyphen /registered /macron
/degree /plusminus /twosuperior /threesuperior
/acute /mu /paragraph /bullet
/cedilla /onesuperior /ordmasculine /guillemotright
/onequarter /onehalf /threequarters /questiondown
/Agrave /Aacute /Acircumflex /Atilde
/Adieresis /Aring /AE /Ccedilla
/Egrave /Eacute /Ecircumflex /Edieresis
/Igrave /Iacute /Icircumflex /Idieresis
/Eth /Ntilde /Ograve /Oacute
/Ocircumflex /Otilde /Odieresis /multiply
/Oslash /Ugrave /Uacute /Ucircumflex
/Udieresis /Yacute /Thorn /germandbls
/agrave /aacute /acircumflex /atilde
/adieresis /aring /ae /ccedilla
/egrave /eacute /ecircumflex /edieresis
/igrave /iacute /icircumflex /idieresis
/eth /ntilde /ograve /oacute
/ocircumflex /otilde /odieresis /divide
/oslash /ugrave /uacute /ucircumflex
/udieresis /yacute /thorn /ydieresis
] def
%%EndResource
%%EndProlog
%%BeginSetup
%%IncludeResource: font Courier-Bold
%%IncludeResource: font Courier
/HFpt_w 10 def
/HFpt_h 10 def
/Courier-Bold /HF-gs-font MF
/HF /HF-gs-font findfont [HFpt_w 0 0 HFpt_h 0 0] makefont def
/Courier /F-gs-font MF
/F-gs-font 10 10 SF
/#copies 1 def
% Pagedevice definitions:
gs_languagelevel 1 gt {
<<
/PageSize [612 792]
>> setpagedevice
} if
/d_page_w 564 def
/d_page_h 744 def
/d_header_x 0 def
/d_header_y 744 def
/d_header_w 564 def
/d_header_h 0 def
/d_footer_x 0 def
/d_footer_y 0 def
/d_footer_w 564 def
/d_footer_h 0 def
/d_output_w 564 def
/d_output_h 744 def
/cols 1 def
%%EndSetup
%%Page: (1) 1
%%BeginPageSetup
_S
24 24 translate
/pagenum 1 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 632 M
(Internet Engineering Task Force E. Kohler) s
5 621 M
(INTERNET-DRAFT UCLA) s
5 610 M
(Intended status: Experimental S. Floyd) s
5 599 M
(Expires: May 2008 ICIR) s
5 588 M
( A. Sathiaseelan) s
5 577 M
( University of Aberdeen) s
5 566 M
( 18 November 2007) s
5 533 M
( Faster Restart for TCP Friendly Rate Control \(TFRC\)) s
5 522 M
( draft-ietf-dccp-tfrc-faster-restart-05.txt) s
5 489 M
(Status of this Memo) s
5 467 M
( By submitting this Internet-Draft, each author represents that any) s
5 456 M
( applicable patent or other IPR claims of which he or she is aware) s
5 445 M
( have been or will be disclosed, and any of which he or she becomes) s
5 434 M
( aware will be disclosed, in accordance with Section 6 of BCP 79.) s
5 412 M
( Internet-Drafts are working documents of the Internet Engineering) s
5 401 M
( Task Force \(IETF\), its areas, and its working groups. Note that) s
5 390 M
( other groups may also distribute working documents as Internet-) s
5 379 M
( Drafts.) s
5 357 M
( Internet-Drafts are draft documents valid for a maximum of six months) s
5 346 M
( and may be updated, replaced, or obsoleted by other documents at any) s
5 335 M
( time. It is inappropriate to use Internet-Drafts as reference) s
5 324 M
( material or to cite them other than as "work in progress.") s
5 302 M
( The list of current Internet-Drafts can be accessed at) s
5 291 M
( http://www.ietf.org/ietf/1id-abstracts.txt.) s
5 269 M
( The list of Internet-Draft Shadow Directories can be accessed at) s
5 258 M
( http://www.ietf.org/shadow.html.) s
5 236 M
( This Internet-Draft will expire on May 2008.) s
5 214 M
(Copyright Notice) s
5 192 M
( Copyright \(C\) The IETF Trust \(2007\).) s
5 104 M
(Kohler, et al. Expires: May 2008 [Page 1]) s
_R
S
%%Page: (2) 2
%%BeginPageSetup
_S
24 24 translate
/pagenum 2 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
(Abstract) s
5 665 M
( TCP-Friendly Rate Control \(TFRC\) is a congestion control mechanism) s
5 654 M
( for unicast flows operating in a best-effort Internet environment.) s
5 643 M
( This document introduces Faster Restart, an optional mechanism for) s
5 632 M
( safely improving the behavior of interactive flows that use TFRC.) s
5 621 M
( Faster Restart is proposed for use with TFRC and with TFRC-SP, the) s
5 610 M
( Small Packet variant of TFRC. We present Faster Restart in general) s
5 599 M
( terms as a congestion control mechanism, and further discuss Faster) s
5 588 M
( Restart for Datagram Congestion Control Protocol \(DCCP\) Congestion) s
5 577 M
( Control IDs 3 and 4.) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 2]) s
_R
S
%%Page: (3) 3
%%BeginPageSetup
_S
24 24 translate
/pagenum 3 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
(Table of Contents) s
5 665 M
( 1. Introduction ....................................................6) s
5 654 M
( 2. Conventions .....................................................9) s
5 643 M
( 3. Faster Restart: Changes to TFRC ................................10) s
5 632 M
( 3.1. Feedback Packets ..........................................10) s
5 621 M
( 3.2. Nofeedback Timer ..........................................13) s
5 610 M
( 4. Faster Restart Discussion ......................................13) s
5 599 M
( 4.1. Worst-Case Scenarios ......................................14) s
5 588 M
( 4.2. Incentives for applications to send unnecessary packets) s
5 577 M
( during idle or data-limited periods ............................15) s
5 566 M
( 4.3. Interoperability Issues ...................................15) s
5 555 M
( 4.3.1. Interoperability Issues with CCID-3 and the RFC) s
5 544 M
( 4342 Errata ...............................................15) s
5 533 M
( 4.4. Faster Restart for TFRC-SP ................................16) s
5 522 M
( 5. Simulations of Faster Restart ..................................16) s
5 511 M
( 6. Implementation Issues ..........................................17) s
5 500 M
( 7. Security Considerations ........................................17) s
5 489 M
( 8. IANA Considerations ............................................17) s
5 478 M
( 9. Thanks .........................................................17) s
5 467 M
( Normative References ..............................................17) s
5 456 M
( Informative References ............................................18) s
5 445 M
( A. Appendix: Simulations ..........................................19) s
5 434 M
( Authors' Addresses ................................................21) s
5 423 M
( Full Copyright Statement ..........................................22) s
5 412 M
( Intellectual Property .............................................22) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 3]) s
_R
S
%%Page: (4) 4
%%BeginPageSetup
_S
24 24 translate
/pagenum 4 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
( NOTE TO RFC EDITOR: PLEASE DELETE THIS NOTE UPON PUBLICATION.) s
5 665 M
( Changes from draft-ietf-dccp-tfrc-faster-restart-04.txt:) s
5 643 M
( * Changed "RTO" to "NFT".) s
5 632 M
( Changed the targeted idle period to the configurable DelayTime.) s
5 621 M
( Feedback from Gerrit Renker.) s
5 599 M
( * Removed Section 4.1 on the receive rate, after it is made) s
5 588 M
( into an Errata for RFC 4342. Feedback from Gerrit Renker.) s
5 566 M
( * General editing from Gorry Fairhurst and Arjuna, and additional) s
5 555 M
( reporting on simulations.) s
5 533 M
( * Added a section on Interoperability Issues.) s
5 511 M
( * Specified CCID 3 and 4 impact in the introduction.) s
5 489 M
( Changes from draft-ietf-dccp-tfrc-faster-restart-03.txt:) s
5 467 M
( * Deleted ping packets, and the section about the implementation) s
5 456 M
( of ping packets in DCCP.) s
5 434 M
( * In Section 3.2, calls to) s
5 423 M
( "Update X_active_recv and X_fast_max;" and) s
5 412 M
( "Interpolate X_fast_max;") s
5 401 M
( had been reversed accidentally. Put them back in the right order.) s
5 379 M
( * Changed Intended Status back to Experimental \(where it started) s
5 368 M
( out\).) s
5 346 M
( * General editing is response to feedback from Gorry.) s
5 324 M
( * Added simulation tests to the list in the section on simulations:) s
5 313 M
( \(1\) simulations) s
5 302 M
( with a worst-case scenario of high congestion, all flows using) s
5 291 M
( TFRC, all flows having various idle times, all flows using Faster) s
5 280 M
( Restart, and variable arrival rates for the TFRC flows \(to create) s
5 269 M
( variable levels of congestion\). And compare this to the same) s
5 258 M
( scenario with no flows using Faster Restart. \(2\) scenarios with) s
5 247 M
( transient changes from routing changes and from variable traffic.) s
5 236 M
( The goal is to explore worse-case scenarios showing off the worst) s
5 225 M
( aspects of Faster Restart.) s
5 203 M
( * Targeted an idle period of at most six minutes, not thirty) s
5 192 M
( minutes. Feedback from Gorry and Ian McDonald.) s
5 170 M
( * Added a section of whether Faster Restart encourages flows to) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 4]) s
_R
S
%%Page: (5) 5
%%BeginPageSetup
_S
24 24 translate
/pagenum 5 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
( pad their sending rate during idle periods.) s
5 665 M
( * Didn't implement suggestion from Lachlan Andrew to decay from) s
5 654 M
( quadrupling to doubling the sending rate gradually. The last) s
5 643 M
( more-than-doubling of the sending rate is probably not a) s
5 632 M
( quadrupling in any case, since the allowed sending rate is) s
5 621 M
( not increased due to quadrupling to more than X_fast_max.) s
5 599 M
( Changes from draft-ietf-dccp-tfrc-faster-restart-02.txt:) s
5 577 M
( * Deleted proposed response to dealing with X_recv for idle or) s
5 566 M
( data-limited periods; RFC3448bis now deals with this instead.) s
5 544 M
( * Deleted the Receive Rate Length option. Also) s
5 533 M
( removed all text about using the inflation factor to) s
5 522 M
( reduce X_recv_in based on the sender's idle time.) s
5 500 M
( * Moved TFRC changes and DCCP-specific changes to separate sections.) s
5 478 M
( * Revised draft to refer to RFC3448bis instead of to RFC3448.) s
5 467 M
( This included modifying sections on "Feedback Packets" and) s
5 456 M
( "Nofeedback Timer".) s
5 434 M
( * Said that CCID 3 could calculate the receive rate only) s
5 423 M
( for one RTT, rather than for longer, after an idle period.) s
5 412 M
( \(When used with RFC3448bis, it shouldn't affect performance) s
5 401 M
( one way or another\).) s
5 379 M
( Changes from draft-ietf-dccp-tfrc-faster-restart-01.txt:) s
5 357 M
( * Added a sentence to Abstract about DCCP.) s
5 335 M
( * Added some text to the Introduction,) s
5 313 M
( * Added sections on "Minimum Sending Rate", "Send Receive) s
5 302 M
( Rate Length Feature", "Nofeedback Timer", and "Simulations) s
5 291 M
( of Faster Restart".) s
5 269 M
( * Added an Appendix on "Simulations".) s
5 247 M
( Changes from draft-ietf-dccp-tfrc-faster-restart-00.txt:) s
5 225 M
( * Added mechanisms for dealing with a more general problem with) s
5 214 M
( idle periods. This includes a section of "Receive Rate) s
5 203 M
( Adjustment".) s
5 181 M
( END OF NOTE TO RFC EDITOR.) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 5]) s
_R
S
%%Page: (6) 6
%%BeginPageSetup
_S
24 24 translate
/pagenum 6 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
(1. Introduction) s
5 665 M
( This document defines congestion control mechanisms that improve the) s
5 654 M
( performance of occasionally idle flows using TCP-Friendly Rate) s
5 643 M
( Control \(TFRC\) [RFC3448] [RFC3448bis]. A data-limited or idle flow) s
5 632 M
( uses less than its allowed sending rate for application-specific) s
5 621 M
( reasons, such as lack of data to send. The responses of Standard) s
5 610 M
( TFRC [RFC3448], and Revised TFRC [RFC3448bis] to long idle or data-) s
5 599 M
( limited periods are summarized in Table 1 below, and the responses of) s
5 588 M
( Standard TCP [RFC2581] and TCP with Congestion Window Validation) s
5 577 M
( [RFC2861] are described in Appendix C of [RFC3448bis]. All of these) s
5 566 M
( mechanisms allow a flow to recover from a long idle period by ramping) s
5 555 M
( up to the allowed sending rate or window. This document specifies) s
5 544 M
( mechanisms that allow TFRC to start at a higher sending rate after an) s
5 533 M
( idle period, and to ramp up faster to the old sending rate after an) s
5 522 M
( idle period.) s
5 500 M
( As this draft is being written, Standard TFRC is specified in) s
5 489 M
( [RFC3448], and TFRC is in the process of being revised, as Revised) s
5 478 M
( TFRC, in [RFC3448bis]. When [RFC3448bis] is approved as a Proposed) s
5 467 M
( Standard document, this draft will be revised, with the phrase) s
5 456 M
( "Standard TFRC" replaced by "Old TFRC", and other language changes as) s
5 445 M
( appropriate.) s
5 423 M
( For Standard TFRC as specified in [RFC3448], a TFRC flow may not send) s
5 412 M
( more than twice X_recv, the rate at which data was received at the) s
5 401 M
( receiver over the previous RTT. Thus in Standard TFRC the previous) s
5 390 M
( receive rate limits the sending rate of applications with highly) s
5 379 M
( variable sending rates, forcing the applications to ramp up, by) s
5 368 M
( doubling their sending rate each round-trip time, from the earlier) s
5 357 M
( data-limited rate to the sending rate allowed by the throughput) s
5 346 M
( equation. TFRC's nofeedback timer halves the allowed sending rate) s
5 335 M
( after each nofeedback timer interval \(at least four round-trip times\)) s
5 324 M
( in which no feedback is received. One result is that applications) s
5 313 M
( must slow-start after being idle for any significant length of time,) s
5 302 M
( in the absence of mechanisms such as Quick-Start [RFC4782] and Quick-) s
5 291 M
( Start for DCCP [GA07].) s
5 269 M
( For Revised TFRC as specified in [RFC3448bis], the previous receive) s
5 258 M
( rate is not used to limit the sending rate during data-limited) s
5 247 M
( periods. Thus, unlike [RFC3448], in [RFC3448bis] applications with) s
5 236 M
( highly variable sending rates are not limited by the previous receive) s
5 225 M
( rates. However, [RFC3448bis] is like [RFC3448] in that the) s
5 214 M
( nofeedback timer is used to halve the allowed sending rate after each) s
5 203 M
( nofeedback timer interval in which no feedback is received. With) s
5 192 M
( [RFC3448] the allowed sending rate is not reduced below two packets) s
5 181 M
( per RTT during idle periods, and with [RFC3448bis] the allowed) s
5 170 M
( sending rate is not reduced below the allowed initial sending rate) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 6]) s
_R
S
%%Page: (7) 7
%%BeginPageSetup
_S
24 24 translate
/pagenum 7 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
( during idle periods.) s
5 665 M
( This behavior is safe, though conservative, for best-effort traffic) s
5 654 M
( in the network. A silent application stops receiving feedback about) s
5 643 M
( the condition of the current network path, and thus should not be) s
5 632 M
( able to send at an arbitrary rate. A data-limited application stops) s
5 621 M
( receiving feedback about whether current network conditions would) s
5 610 M
( support higher rates. However, this behavior also affects the) s
5 599 M
( perceived performance of interactive applications such as voice.) s
5 588 M
( Connections for interactive telephony and conference applications,) s
5 577 M
( for example, will usually have one party active at a time, with) s
5 566 M
( seamless switching between active parties. TFRC's reduction of the) s
5 555 M
( allowed sending rate, and slow-starting back to a higher sending) s
5 544 M
( rate, after every switch between parties could seriously degrade) s
5 533 M
( perceived performance. Some of the strategies suggested for coping) s
5 522 M
( with this problem, such as sending padding data during application) s
5 511 M
( idle periods, might have worse effects on the network than simply) s
5 500 M
( switching onto the desired rate with no slow-start.) s
5 478 M
( There is some justification for somewhat accelerating the slow start) s
5 467 M
( process after idle periods, as opposed to at the beginning of a) s
5 456 M
( connection. A flow that fairly achieves a sending rate of X has) s
5 445 M
( proved, at least, that some path between the endpoints can support) s
5 434 M
( that rate. The path might change, due to endpoint reset or routing) s
5 423 M
( adjustments; or many new connections might start up, significantly) s
5 412 M
( reducing the application's fair rate. However, it seems reasonable) s
5 401 M
( to allow an application to possibly contribute to limited transient) s
5 390 M
( congestion in times of change, in return for improving application) s
5 379 M
( responsiveness.) s
5 357 M
( This document suggests a relatively simple approach to this problem.) s
5 346 M
( Standard TFRC [RFC3448] specifies that the allowed sending rate is) s
5 335 M
( never reduced below two packets per RTT as the result of a nofeedback) s
5 324 M
( timer after an idle period. Following [RFC3390], CCID-3 [RFC4342]) s
5 313 M
( and Revised TFRC [RFC3448bis] specify that the allowed sending rate) s
5 302 M
( is never reduced below the TCP initial sending rate of two or four) s
5 291 M
( packets per RTT, depending on packet size, as the result of a) s
5 280 M
( nofeedback timer after an idle period. Faster Restart doubles this) s
5 269 M
( allowed sending rate after idle periods. Thus, the sending rate) s
5 258 M
( after an idle period is not reduced below a rate Y between four and) s
5 247 M
( eight packets per RTT, depending on the packet size. The rate Y is) s
5 236 M
( restricted to at most 8760 bytes per RTT \(which is twice TCP's) s
5 225 M
( maximum allowed initial window size\).) s
5 203 M
( In addition, because flows already have some \(possibly old\)) s
5 192 M
( information about the path, Faster Restart allows flows to quadruple) s
5 181 M
( their sending rate in every congestion-free RTT, instead of doubling,) s
5 170 M
( upwards towards the previously achieved rate. When the TFRC sender) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 7]) s
_R
S
%%Page: (8) 8
%%BeginPageSetup
_S
24 24 translate
/pagenum 8 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
( detects congestion, the sender leaves Faster Restart and changes into) s
5 676 M
( congestion avoidance. These changes are summarized in the table) s
5 665 M
( below. In this document, "NFT" refers to the NoFeedback Timer) s
5 654 M
( interval for TFRC; this is roughly equivalent to the Retransmit) s
5 643 M
( TimeOut \(RTO\) interval for TCP.) s
5 610 M
( ------------------------------------------------------------------) s
5 599 M
( - Standard TFRC -) s
5 588 M
( ------------------------------------------------------------------) s
5 577 M
( Idle period:) s
5 566 M
( Halve allowed sending rate each NFT, not below two packets per RTT.) s
5 555 M
( After sending again, double the sending rate each RTT.) s
5 544 M
( Application-limited period:) s
5 533 M
( Send at most twice X_recv.) s
5 522 M
( As a result, at most double the sending rate each RTT.) s
5 511 M
( ------------------------------------------------------------------) s
5 489 M
( ------------------------------------------------------------------) s
5 478 M
( - Revised TFRC -) s
5 467 M
( ------------------------------------------------------------------) s
5 456 M
( Idle period:) s
5 445 M
( Halve allowed sending rate each NFT, not below initial sending rate.) s
5 434 M
( After sending again, double the sending rate each RTT.) s
5 423 M
( Application-limited period:) s
5 412 M
( Sending rate not limited by X_recv.) s
5 401 M
( ------------------------------------------------------------------) s
5 379 M
( ------------------------------------------------------------------) s
5 368 M
( - Revised TFRC with Faster Restart -) s
5 357 M
( ------------------------------------------------------------------) s
5 346 M
( Idle period:) s
5 335 M
( Halve allowed sending rate each NFT, not below twice initial rate.) s
5 324 M
( \(Specified in Section 3.2.\)) s
5 313 M
( After sending again, quadruple the sending rate towards old rate.) s
5 302 M
( \(Specified in Section 3.1.\)) s
5 291 M
( Application-limited period:) s
5 280 M
( Sending rate not limited by X_recv.) s
5 269 M
( ------------------------------------------------------------------) s
5 247 M
( Table 1: Behavior of TFRC, with and without Faster Restart.) s
5 225 M
( The congestion control mechanisms defined here are intended to apply) s
5 214 M
( to any implementations of TFRC, including that in DCCP's CCID 3 and) s
5 203 M
( CCID 4 [RFC4342], [CCID4]. These mechanisms change only CCID 3 and 4) s
5 192 M
( sender behavior and do not change DCCP packets in externally visible) s
5 181 M
( ways \(except in that the sending rate will be higher after an idle) s
5 170 M
( period\). This reduces interoperability concerns. Any DCCP CCID 3) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 8]) s
_R
S
%%Page: (9) 9
%%BeginPageSetup
_S
24 24 translate
/pagenum 9 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
( or 4 sender MAY therefore use Faster Restart algorithms at its) s
5 676 M
( discretion, without negotiation with the corresponding receiver.) s
5 654 M
( While we also believe that TCP could safely use a similar Faster) s
5 643 M
( Restart mechanism, we do not specify it here. Our assumption is that) s
5 632 M
( flows that are sensitive to restrictions to the sending rate after) s
5 621 M
( idle periods are more likely to use TFRC than to use TCP or TCP-like) s
5 610 M
( congestion control.) s
5 588 M
(2. Conventions) s
5 566 M
( The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",) s
5 555 M
( "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this) s
5 544 M
( document are to be interpreted as described in [RFC2119].) s
5 522 M
( The Faster Restart mechanism refers to several existing TFRC state) s
5 511 M
( variables, including the following:) s
5 489 M
( R: The RTT estimate.) s
5 467 M
( X: The current allowed sending rate in bytes per second.) s
5 445 M
( p: The recent loss event rate.) s
5 423 M
( X_recv:) s
5 412 M
( The rate at which the receiver estimates that data was received) s
5 401 M
( since the last feedback report was sent.) s
5 379 M
( s: The packet size in bytes.) s
5 357 M
( Faster Restart uses the following variable from [RFC3448bis]:) s
5 335 M
( recv_limit:) s
5 324 M
( The limit on the sending rate that is computed from the receive) s
5 313 M
( rate.) s
5 291 M
( Faster Restart also introduces new state variables to TFRC, as) s
5 280 M
( follows:) s
5 258 M
( X_active_recv:) s
5 247 M
( The receiver's estimated receive rate reported during a recent) s
5 236 M
( active sending period. An active sending period is a period in) s
5 225 M
( which the sender has not experienced a loss event. X_active_recv) s
5 214 M
( is initialized to 0 until there has been an active sending period,) s
5 203 M
( and X_active_recv is reduced after a loss event.) s
5 181 M
( T_active_recv:) s
5 170 M
( The time at which X_active_recv was measured. T_active_recv is) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 9]) s
_R
S
%%Page: (10) 10
%%BeginPageSetup
_S
24 24 translate
/pagenum 10 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
( initialized to the start time of the connection.) s
5 665 M
( recover_rate:) s
5 654 M
( The minimum restart rate allowed by Faster Restart after an idle) s
5 643 M
( period. Note that Faster Restart flows can drop below this rate) s
5 632 M
( as the result of experienced congestion \(e.g. actual loss) s
5 621 M
( feedback\). Recover_rate is defined as follows:) s
5 599 M
( recover_rate = min\(8*s, max\(4*s, 8760 bytes\)\)/R.) s
5 577 M
( Faster Restart also uses the following, which could be implemented as) s
5 566 M
( a temporary variable:) s
5 544 M
( X_fast_max:) s
5 533 M
( The rate at which the sender should stop quadrupling its sending) s
5 522 M
( rate, and return to at most doubling its sending rate.) s
5 500 M
( Other variables have values as described in [RFC3448] and) s
5 489 M
( [RFC3448bis].) s
5 467 M
(3. Faster Restart: Changes to TFRC) s
5 445 M
(3.1. Feedback Packets) s
5 423 M
( The Faster Restart algorithm replaces the line:) s
5 401 M
( recv_limit = 2 * max \(X_recv_set\);) s
5 379 M
( in step \(4\) of Section 4.3, "Sender Behavior When a Feedback Packet) s
5 368 M
( is Received", of [RFC3448bis]. This line specifies the limitation on) s
5 357 M
( the sending rate calculated based on the recent receive rate, and in) s
5 346 M
( [RFC3448bis] allows the sender to slow-start back up to the previous) s
5 335 M
( sending rate after an idle period, doubling its sending rate after) s
5 324 M
( each round-trip time.) s
5 302 M
( This document replaces the line above so that during recovery from an) s
5 291 M
( idle period, the TFRC sender can quadruple its sending rate each) s
5 280 M
( \(congestion-free\) round-trip time, until it reaches its old sending) s
5 269 M
( rate before the idle period. This modification uses three new) s
5 258 M
( variables: X_active_recv specifies the maximum receive rate achieved) s
5 247 M
( before the idle period, T_active_recv specifies the time of the last) s
5 236 M
( update of X_active_recv, and X_fast_max specifies the adjusted rate) s
5 225 M
( at which the sender should stop quadrupling its sending rate and) s
5 214 M
( continue to its default behavior of doubling its sending rate.) s
5 192 M
( The procedure "Update X_active_recv and X_fast_max" below increases) s
5 181 M
( the two variables in response to increases in the reported receive) s
5 170 M
( rate and reduces them after a report of a lost packet or an) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 10]) s
_R
S
%%Page: (11) 11
%%BeginPageSetup
_S
24 24 translate
/pagenum 11 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
( indication of congestion \(e.g. an ECN-marked packet\).) s
5 665 M
( Update X_active_recv and X_fast_max:) s
5 654 M
( If \(the feedback packet does not indicate a loss or mark,) s
5 643 M
( and X_recv >= X_fast_max\)) s
5 632 M
( X_active_recv = X_fast_max = X_recv,) s
5 621 M
( T_active_recv = current time.) s
5 610 M
( Else if \(the feedback packet DOES indicate a loss or mark,) s
5 599 M
( and X_recv < X_fast_max\)) s
5 588 M
( X_active_recv = X_fast_max = X_recv/2,) s
5 577 M
( T_active_recv = current time.) s
5 555 M
( The parameter X_active_recv gives an upper bound on the rate) s
5 544 M
( achievable through Faster Restart, and is only modified by the) s
5 533 M
( "Update X_active_rate and X_fast_max" procedure. This modification) s
5 522 M
( is based on the contents of the feedback packet and the value of) s
5 511 M
( X_fast_max. X_active_recv is updated as the connection achieves) s
5 500 M
( higher congestion-free transmit rates. X_active_recv is reduced on) s
5 489 M
( congestion feedback, to prevent an inappropriate Faster Restart until) s
5 478 M
( a new stable active rate is achieved. Specifically, when congestion) s
5 467 M
( feedback is received at a low sending rate, the sender reduces) s
5 456 M
( X_active_recv to X_recv/2, allowing a limited Faster Restart up to a) s
5 445 M
( likely-safe rate.) s
5 423 M
( For some transport protocols using TFRC, the feedback packets might) s
5 412 M
( report the loss event rate, but not explicitly report lost or marked) s
5 401 M
( packets. For such protocols, the sender in the "Update X_active_rate) s
5 390 M
( and X_fast_max" procedure can infer that a feedback packet indicates) s
5 379 M
( a loss or mark by looking at the reported loss event rate. If the) s
5 368 M
( current or previous feedback packet reported an increase in the loss) s
5 357 M
( event rate, then the current feedback packet is assumed to indicate a) s
5 346 M
( loss or mark. \(If the previous feedback packet reported an increase) s
5 335 M
( in the loss event rate, then a loss event began in the interval) s
5 324 M
( covered by that feedback packet. However, the loss event can cover) s
5 313 M
( up to a round-trip time of data, so the second half of the loss) s
5 302 M
( event, including additional lost or marked packets, could be covered) s
5 291 M
( by the second feedback packet.\)) s
5 269 M
( The "Interpolate X_fast_max" procedure determines X_fast_max, the) s
5 258 M
( adjusted rate at which Faster Restart should stop. The procedure) s
5 247 M
( sets X_fast_max to something between zero and X_active_recv,) s
5 236 M
( depending on the time since X_active_recv was last updated. The) s
5 225 M
( procedure allows full Faster Restart up to the old sending rate) s
5 214 M
( X_active_recv after a short idle period, but requires more) s
5 203 M
( conservative behavior after a longer idle period. Thus, if at most) s
5 192 M
( DecayTime has elapsed since the last update of X_active_recv, for a) s
5 181 M
( default DecayTime of two minutes, then X_fast_max is set to) s
5 170 M
( X_active_recv. If 3*DecayTime or more has elapsed, X_fast_max is set) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 11]) s
_R
S
%%Page: (12) 12
%%BeginPageSetup
_S
24 24 translate
/pagenum 12 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
( to zero. Linear interpolation is used between these extremes.) s
5 665 M
( The default DecayTime of two minutes is chosen to strike a balance) s
5 654 M
( between the needs of applications, and the time intervals over which) s
5 643 M
( connections might reasonably quadruple back up to their old sending) s
5 632 M
( rates after idle periods. In terms of the needs of applications,) s
5 621 M
( models of voice traffic generally use average idle times between 0.5) s
5 610 M
( and two seconds [JS00] \(Section 3\). However, in terms of changes in) s
5 599 M
( path characteristics, Faster Restart does not assume that the) s
5 588 M
( previous sending rate is valid after an idle period; Faster Restart) s
5 577 M
( simply assumes that a connection may *quadruple* rather than *double*) s
5 566 M
( its sending rate up to the previous rate. Therefore, while an overly) s
5 555 M
( long DecayTime is not likely to lead to congestion collapse, it could) s
5 544 M
( result in unnecessary packet drops, and therefore in reduced) s
5 533 M
( performance for the application itself. Path congestion levels can) s
5 522 M
( change over time scales of round-trip times, which are generally) s
5 511 M
( between 10 and several hundred milliseconds; more dramatic changes in) s
5 500 M
( path characteristics \(e.g., routing changes, changes in link) s
5 489 M
( bandwidth\) happen less frequently. For now, the DecayTime may be a) s
5 478 M
( configurable parameter. Future work may shed more light on optimum) s
5 467 M
( values for DecayTime.) s
5 445 M
( Interpolate X_fast_max:) s
5 434 M
( // If achieved X_active_recv <= 1 minute ago,) s
5 423 M
( // set X_fast_max to X_active_recv;) s
5 412 M
( // If achieved X_active_recv >= 3 minutes ago,) s
5 401 M
( // set X_fast_max to zero;) s
5 390 M
( // If in between, interpolate.) s
5 379 M
( delta_T = now - T_active_recv;) s
5 368 M
( F = \(6 min - min\(max\(delta_T, 2 min\), 6 min\)\) / \(2 min\);) s
5 357 M
( X_fast_max = F * X_active_recv;) s
5 335 M
( The pseudocode above uses the temporary variables delta_T and F.) s
5 313 M
( Faster Restart replaces the following line from step \(4\) of Section) s
5 302 M
( 4.3 of [RFC3448bis]:) s
5 280 M
( recv_limit = 2 * max \(X_recv_set\);) s
5 258 M
( with the following:) s
5 236 M
( Interpolate X_fast_max;) s
5 225 M
( Update X_active_recv and X_fast_max;) s
5 214 M
( recv_limit = 2 * max \(X_recv_set\);) s
5 203 M
( If \(recv_limit < X_fast_max\)) s
5 192 M
( recv_limit = min\(2*recv_limit, X_fast_max\);) s
5 170 M
( In summary, when a feedback packet is received, as specified in) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 12]) s
_R
S
%%Page: (13) 13
%%BeginPageSetup
_S
24 24 translate
/pagenum 13 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
( [RFC3448bis], then the sender updates the round-trip time estimate) s
5 676 M
( and the NFT \(NoFeedback Timer\), and updates X_recv_set, the set of) s
5 665 M
( recent X_recv values, and then executes the procedure above.) s
5 654 M
( X_fast_max always represents the interpolated value from highest) s
5 643 M
( X_recv reported since the last loss event. However, because) s
5 632 M
( X_recv_set contains only X_recv values from the most recent two) s
5 621 M
( round-trip times, the calculated recv_limit could be less than) s
5 610 M
( X_fast_max. In this case, recv_limit is doubled, up to at most) s
5 599 M
( X_fast_max. Faster Restart's doubling of recv_limit allows the TFRC) s
5 588 M
( sender to quadruple its sending rate each round-trip time after an) s
5 577 M
( idle period.) s
5 555 M
(3.2. Nofeedback Timer) s
5 533 M
( Section 4.4 of [RFC3448bis] specifies when the allowed sending rate) s
5 522 M
( is halved after the nofeedback timer expires. In particular,) s
5 511 M
( [RFC3448bis] specifies that if the sender has been idle since the) s
5 500 M
( nofeedback timer was set, then the allowed sending rate is not) s
5 489 M
( reduced below recover_rate, which in [RFC3448bis] is set to the) s
5 478 M
( initial_rate of W_init/R, for:) s
5 456 M
( W_init = min\(4*s, max\(2*s, 4380\)\),) s
5 434 M
( for segment size s. In contrast, this document sets recover_rate to) s
5 423 M
( twice the initial_rate, as follows:) s
5 401 M
( recover_rate = 2*W_init/R;) s
5 368 M
(4. Faster Restart Discussion) s
5 346 M
( Standard TCP has historically dealt with idleness and data-limited) s
5 335 M
( flows either by keeping cwnd entirely open \("immediate start"\) or by) s
5 324 M
( entering slow-start, as recommended in RFC 2581 in response to an) s
5 313 M
( idle period. The first option is too liberal, the second too) s
5 302 M
( conservative. Clearly a short idle or data-limited period is not a) s
5 291 M
( new connection: the sending rate maintained before the idle or data-) s
5 280 M
( limited period shows that previously, the connection could fairly) s
5 269 M
( sustain some rate without adversely impacting other flows. However,) s
5 258 M
( longer idle periods are more problematic. Idle periods of many) s
5 247 M
( minutes would seem to require slow-start.) s
5 225 M
( RFC 2861 [RFC2861] gives a moderate mechanism for TCP, where the) s
5 214 M
( congestion window is halved for every retransmit timeout interval) s
5 203 M
( that the sender has remained idle, down to the initial window, and) s
5 192 M
( the window is re-opened in slow-start when the idle period is over.) s
5 181 M
( TFRC in [RFC3448bis] roughly follows [RFC2861] for the response to an) s
5 170 M
( idle period. Unlike [RFC2861], however, [RFC3448bis] follows) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 13]) s
_R
S
%%Page: (14) 14
%%BeginPageSetup
_S
24 24 translate
/pagenum 14 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
( Standard TCP in its responses to a data-limited period, and does not) s
5 676 M
( reduce the allowed sending rate in response to data-limited periods.) s
5 654 M
(4.1. Worst-Case Scenarios) s
5 632 M
( Faster Restart should be acceptable for TFRC if its worst-case) s
5 621 M
( scenarios are acceptable. Realistic worst-case scenarios might) s
5 610 M
( include the following scenarios:) s
5 588 M
( o Path changes: The path changes and the old rate is not acceptable) s
5 577 M
( on the new path. RTTs are shorter on the new path too, so Faster) s
5 566 M
( Restart takes bandwidth from other connections for multiple RTTs,) s
5 555 M
( not just one. \(This can happen with TCP or with TFRC without) s
5 544 M
( Faster Restart, but Faster Restart could make this behavior more) s
5 533 M
( severe.\)) s
5 511 M
( o Synchronized flows: Several connections enter Faster Restart) s
5 500 M
( simultaneously. If the path is congested, the extra load) s
5 489 M
( resulting from Faster Restart could be twice as bad as the extra) s
5 478 M
( load if the connections had simply slow-started from their allowed) s
5 467 M
( initial sending rate.) s
5 445 M
( o Many forms of burstiness: Variable-rate connections using Faster) s
5 434 M
( Restart share the congested link with short TCP or DCCP) s
5 423 M
( connections starting and stopping, with initial windows of three) s
5 412 M
( or four packets. The aggregate traffic could also include TCP) s
5 401 M
( connections with short quiescent periods \(e.g., web browsing) s
5 390 M
( sessions using HTTP 1.1\), or bursty higher-priority traffic. As a) s
5 379 M
( result of the bursty traffic, the aggregate arrival rate varies) s
5 368 M
( from one RTT to the next. The transient congestion will be) s
5 357 M
( particularly severe if the congested link is an access link) s
5 346 M
( instead of a backbone link; the level of statistical multiplexing) s
5 335 M
( on an access link may not be sufficiently high to "smooth out" the) s
5 324 M
( burstiness.) s
5 302 M
( o Wireless links: The network allocates capacity based on traffic) s
5 291 M
( conditions, as in some current wireless technologies, such as) s
5 280 M
( Bandwidth on Demand \(BoD\) links [RFC3819] where capacity is) s
5 269 M
( variable and dependent on several parameters other than network) s
5 258 M
( congestion. In this case, the old sending rate might not be) s
5 247 M
( acceptable after a change in capacity for the wireless link during) s
5 236 M
( an idle period.) s
5 214 M
( Further analysis is required to analyze the effects of these) s
5 203 M
( scenarios.) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 14]) s
_R
S
%%Page: (15) 15
%%BeginPageSetup
_S
24 24 translate
/pagenum 15 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
(4.2. Incentives for applications to send unnecessary packets during) s
5 676 M
(idle or data-limited periods) s
5 654 M
( How does Faster Restart affect an application's incentive to pad its) s
5 643 M
( sending rate by sending unnecessary packets during idle or data-) s
5 632 M
( limited periods? We would like to limit an application's incentive) s
5 621 M
( to pad its sending rate during idle or data-limited periods; if all) s
5 610 M
( applications were to pad their sending rates, it could reduce the) s
5 599 M
( available bandwidth, and degrade the performance for all flows on the) s
5 588 M
( congested link.) s
5 566 M
( With Standard TFRC as specified in [RFC3448], a data-limited TFRC) s
5 555 M
( flow may not send more than twice X_recv, the rate at which data was) s
5 544 M
( received at the receiver over the previous RTT. Thus, with Standard) s
5 533 M
( TFRC, one could argue that a variable-rate application over an) s
5 522 M
( uncongested path does have some incentive to pad its sending rate.) s
5 500 M
( With Revised TFRC as specified in [RFC3448bis], the allowed sending) s
5 489 M
( rate after an idle period is larger than the allowed sending rate) s
5 478 M
( with Standard TFRC. Further, with Revised TFRC the receive rate) s
5 467 M
( reported in feedback packets is not used to limit the sending rate) s
5 456 M
( during data-limited periods. Thus, with Revised TFRC an application) s
5 445 M
( has less incentive to pad its sending rate than with Standard TFRC.) s
5 434 M
( However, with Revised TFRC an application could have some incentive) s
5 423 M
( to pad its sending rate just enough to maintain the status of "data-) s
5 412 M
( limited" instead of "idle", by sending at least one packet every four) s
5 401 M
( round-trip times.) s
5 379 M
( By allowing TFRC to revert to its old sending rate more quickly after) s
5 368 M
( an idle period, Faster Restart could reduce an application's) s
5 357 M
( incentive to pad its sending rate.) s
5 335 M
(4.3. Interoperability Issues) s
5 313 M
( Faster Restart is a sender-side only modification to TFRC, and is) s
5 302 M
( intended to work with any TFRC receiver using the same transport) s
5 291 M
( protocol. The current standard for TFRC is RFC 3448. After) s
5 280 M
( [RFC3448bis] is standardized, the authors of this document will) s
5 269 M
( verify that Faster Restart works with either an RFC3448 or an) s
5 258 M
( RFC3448bis receiver.) s
5 236 M
(4.3.1. Interoperability Issues with CCID-3 and the RFC 4342 Errata) s
5 214 M
( For the particular case of TFRC as used in CCID-3 or CCID-4 in DCCP,) s
5 203 M
( there are currently two variants of CCID-3 receivers. For TFRC as) s
5 192 M
( specified in [RFC3448], the receiver reports the receive rate) s
5 181 M
( measured over the most recent round-trip time. In contrast, for) s
5 170 M
( CCID-3 as specified in [RFC4342], the receiver reports the receive) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 15]) s
_R
S
%%Page: (16) 16
%%BeginPageSetup
_S
24 24 translate
/pagenum 16 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
( rate measured over the interval since the last feedback packet was) s
5 676 M
( received. These two methods can differ for feedback packets sent) s
5 665 M
( after a loss event or after an idle period. To correct this, the RFC) s
5 654 M
( 4342 Errata [RFC4342Errat] now specifies that the receiver reports) s
5 643 M
( the receive rate measured over the most recent round-trip time, as in) s
5 632 M
( RFC 3448.) s
5 610 M
( Because Faster Restart is being specified only for a sender using) s
5 599 M
( [RFC3448bis], and not for a sender using [RFC3448], Faster Restart in) s
5 588 M
( CCID-3 should interoperate with a CCID-3 receiver as specified in) s
5 577 M
( [RFC4342], with a CCID-3 receiver as specified in [RFC4342] and) s
5 566 M
( updated by the RFC 4342 Errata, or with a CCID-3 receiver as) s
5 555 M
( specified in [RFC4342] updated by both the RFC 4342 Errata and by) s
5 544 M
( [RFC3448bis]. In particular, with Faster Restart in CCID-3 \(or) s
5 533 M
( CCID-4\) with RFC3448bis, the sender's sending rate is not limited by) s
5 522 M
( the first feedback packet received after an idle period, so Faster) s
5 511 M
( Restart should perform well even with a CCID-3 \(or CCID-4\) receiver) s
5 500 M
( following RFC 4342 and not updated by the RFC 4342 Errata.) s
5 478 M
(4.4. Faster Restart for TFRC-SP) s
5 456 M
( We note that Faster Restart with TFRC-SP [RFC4828] is considerably) s
5 445 M
( more restrained than Faster Restart with TFRC. In TFRC-SP, the) s
5 434 M
( sender is restricted to sending at most one packet every Min) s
5 423 M
( Interval.) s
5 401 M
(5. Simulations of Faster Restart) s
5 379 M
( Some test case scenarios based on simulation analysis are described) s
5 368 M
( in Appendix A. These simulations follow the guidelines set in) s
5 357 M
( [RFC4828]. These are:) s
5 335 M
( 1. Fairness to standard TCP and TFRC: The simulation tests examine) s
5 324 M
( whether flows that use Faster Restart allow TCP and TFRC flows can) s
5 313 M
( achieve their share of the path capacity.) s
5 291 M
( 2. Fairness within Faster Restart: The simulation tests examine how) s
5 280 M
( multiple competing Faster Restart flows share the available) s
5 269 M
( capacity among them.) s
5 247 M
( 3. Response to transient events: The simulation tests examine how a) s
5 236 M
( Faster Restart flow reacts to a sudden congestion event.) s
5 214 M
( 4. Behavior in a range of environments: Tests assess a range of) s
5 203 M
( bandwidths, RTTs, and varying idle periods.) s
5 181 M
( A set of initial simulation results are described in [S07]. We note) s
5 170 M
( some of the important results here.) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 16]) s
_R
S
%%Page: (17) 17
%%BeginPageSetup
_S
24 24 translate
/pagenum 17 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
( o Faster Restart does improve the performance of a flow after an) s
5 676 M
( idle period by faster restarting when compared to TFRC. The) s
5 665 M
( results indicate that the worst case packet delay distribution is) s
5 654 M
( small for Faster Restart than for TFRC.) s
5 632 M
( o The effect of Faster Restart restarting after an idle period seems) s
5 621 M
( to have an effect on other competing flows only when the Faster) s
5 610 M
( Restart flow has a high sending rate before it enters the idle) s
5 599 M
( period.) s
5 577 M
( o When the Faster Restart flows experience losses and hence reduce) s
5 566 M
( their rates to a lower rate prior to entering an idle period, the) s
5 555 M
( effect of faster restarting is similar to that of slow-start.) s
5 533 M
( A later version of this draft will provide more discussion on these) s
5 522 M
( results in the appendix and implications will be noted here.) s
5 500 M
(6. Implementation Issues) s
5 478 M
( TBA) s
5 456 M
(7. Security Considerations) s
5 434 M
( TRFC security considerations are discussed in [RFC3448]. DCCP) s
5 423 M
( security considerations are discussed in [RFC4340]. Faster Restart) s
5 412 M
( adds no additional security considerations.) s
5 390 M
(8. IANA Considerations) s
5 368 M
( There are no IANA considerations.) s
5 346 M
(9. Thanks) s
5 324 M
( We thank the DCCP Working Group for feedback and discussions; we) s
5 313 M
( particularly thank Gorry Fairhurst. We thank Vlad Balan and Gerrit) s
5 302 M
( Renker for pointing out problems with the mechanisms discussed in) s
5 291 M
( previous versions of the draft.) s
5 258 M
(Normative References) s
5 236 M
( [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate) s
5 225 M
( Requirement Levels", BCP 14, RFC 2119, March 1997.) s
5 203 M
( [RFC3448] Handley, M., Floyd, S., Padhye, J., and J. Widmer,) s
5 192 M
( "TCP Friendly Rate Control \(TFRC\): Protocol) s
5 181 M
( Specification", RFC 3448, Proposed Standard, January) s
5 170 M
( 2003.) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 17]) s
_R
S
%%Page: (18) 18
%%BeginPageSetup
_S
24 24 translate
/pagenum 18 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
( [RFC3448bis] Handley, M., Floyd, S., Padhye, J., and J. Widmer,) s
5 676 M
( "TCP Friendly Rate Control \(TFRC\): Protocol) s
5 665 M
( Specification", internet draft draft-ietf-dccp-) s
5 654 M
( rfc3448bis-02.txt, work-in-progress, July 2007.) s
5 632 M
( [RFC4340] Kohler, E., Handley, M., and S. Floyd, "Datagram) s
5 621 M
( Congestion Control Protocol \(DCCP\)", RFC 4340, March) s
5 610 M
( 2006.) s
5 588 M
( [RFC4342] Floyd, S., Kohler, E., and J. Padhye, "Profile for) s
5 577 M
( Datagram Congestion Control Protocol \(DCCP\) Congestion) s
5 566 M
( Control ID 3: TCP-Friendly Rate Control \(TFRC\)", RFC) s
5 555 M
( 4342, March 2006.) s
5 533 M
(Informative References) s
5 511 M
( [CCID4] Floyd, S., and E. Kohler, "Profile for Datagram) s
5 500 M
( Congestion Control Protocol \(DCCP\) Congestion ID 4:) s
5 489 M
( TCP-Friendly Rate Control for Small Packets \(TFRC-) s
5 478 M
( SP\)", Internet-Draft draft-ietf-dccp-ccid4-00.txt,) s
5 467 M
( work in progress, October 2007.) s
5 445 M
( [GA07] "Quick-Start for the Datagram Congestion Control) s
5 434 M
( Protocol \(DCCP\)", Internet-Draft draft-fairhurst-) s
5 423 M
( tsvwg-dccp-qs-02.txt, work in progress, November 2007.) s
5 401 M
( [JS00] W. Jiang and H. Schulzrinne, Analysis of On-Off) s
5 390 M
( Patterns in VoIP and Their Effect on Voice Traffic) s
5 379 M
( Aggregation, Proceedings of the Ninth Conference on) s
5 368 M
( Computer Communications and Networks \(ICCCN\), October) s
5 357 M
( 2000.) s
5 335 M
( [RFC2581] Allman, M., Paxson, V., and W. Stevens, "TCP) s
5 324 M
( Congestion Control", RFC 2581, April 1999.) s
5 302 M
( [RFC2861] Handley, M., Padhye, J., and S. Floyd, "TCP Congestion) s
5 291 M
( Window Validation", RFC 2861, June 2000.) s
5 269 M
( [RFC3390] Allman, M., Floyd, S., and C. Partridge, "Increasing) s
5 258 M
( TCP's Initial Window", RFC 3390, October 2002.) s
5 236 M
( [RFC3819] Karn, P., Ed., Bormann, C., Fairhurst, G., Grossman,) s
5 225 M
( D., Ludwig, R., Mahdavi, J., Montenegro, G., Touch,) s
5 214 M
( J., and L. Wood, "Advice for Internet Subnetwork) s
5 203 M
( Designers", RFC 3819, July 2004.) s
5 181 M
( [RFC4342Errat] RFC Errata for RFC 4342, URL "http://www.rfc-) s
5 170 M
( editor.org/errata.php".) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 18]) s
_R
S
%%Page: (19) 19
%%BeginPageSetup
_S
24 24 translate
/pagenum 19 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
( [RFC4782] Floyd, S., Allman, M., Jain, A., and P. Sarolahti,) s
5 676 M
( "Quick-Start for TCP and IP", RFC 4782, June 2006.) s
5 654 M
( [RFC4828] Floyd, S., and E. Kohler, "TCP Friendly Rate Control) s
5 643 M
( \(TFRC\): the Small-Packet \(SP\) Variant", RFC 4828,) s
5 632 M
( April 2007.) s
5 610 M
( [S07] Sathiaseelan, A., Faster Restart - Analysis, URL) s
5 599 M
( www.erg.abdn.ac.uk/users/arjuna/faster-restart.pdf.) s
5 577 M
(A. Appendix: Simulations) s
5 555 M
( This appendix describes a set of initial test case scenarios for) s
5 544 M
( simulation analysis of Faster Restart. The simulation results were) s
5 533 M
( performed using the ns-2 simulator. The topology is the classic) s
5 522 M
( dumb-bell topology used in many simulations of TCP. The bottleneck) s
5 511 M
( capacity was set to 6 Mbps. We considered various link delays. The) s
5 500 M
( bottleneck queue was set to the bandwidth delay product for all the) s
5 489 M
( simulations. The results presented here were based on an average of) s
5 478 M
( 20 simulation runs.) s
5 456 M
( Several types of flows are considered:) s
5 434 M
( o Bulk TCP Flows.) s
5 412 M
( o Interactive \(short\) TCP Flows.) s
5 390 M
( o TFRC Flows with and without Faster Restart.) s
5 368 M
( o TFRC-SP Flows with and without Faster Restart.) s
5 346 M
( The implications on other flows \(e.g. using UDP\) may be extrapolated) s
5 335 M
( from this.) s
5 313 M
( For these simulations, we consider two application rates.) s
5 291 M
( o Small media flows: These have a similar rate to voice over IP) s
5 280 M
( with a media bit rate of 64 Kbps \(using segments of 160 bytes and) s
5 269 M
( a nominal transmit rate of 8 KBps\).) s
5 247 M
( o Large media flows: These have a similar rate to medium quality) s
5 236 M
( video over IP with a media bit rate of 512 Kbps \(using segments of) s
5 225 M
( size 1000 bytes and a nominal transmit rate of 64 KBps\).) s
5 203 M
( The transmit buffer was set to zero packets unless otherwise noted.) s
5 192 M
( This means there would no buffering between the application and the) s
5 181 M
( transport protocol.) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 19]) s
_R
S
%%Page: (20) 20
%%BeginPageSetup
_S
24 24 translate
/pagenum 20 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
( The simulations model the effect of an idle period in which the) s
5 676 M
( application does not attempt to send any data for a period of time,) s
5 665 M
( then resumes transmission. Various idle times are considered in the) s
5 654 M
( simulation experiments.) s
5 632 M
( The simulation scenarios include the following. These are intended) s
5 621 M
( to be illustrative, rather than exact models of the application) s
5 610 M
( behavior.) s
5 588 M
( o Performance of a long-lived \(bulk\) TCP flow \(e.g. FTP\) with TFRC) s
5 577 M
( flows \(with and without Faster Restart\): The test scenario would) s
5 566 M
( involve a single large FTP flow with varying number of large media) s
5 555 M
( flows. Each large media flow becomes idle for one second and then) s
5 544 M
( restarts. The FTP flow starts during the idle period. The) s
5 533 M
( throughput performance of the single FTP flow would be plotted for) s
5 522 M
( varying number of large media flows. Does the single FTP flow get) s
5 511 M
( at least 1/n share of the bandwidth, where TFRC flows decrease the) s
5 500 M
( bandwidth received by the TCP flow?) s
5 478 M
( o Performance of small TCP flows \(HTTP\) with TFRC flows with and) s
5 467 M
( without Faster Restart: The test scenario would involve a single) s
5 456 M
( large media flow which runs for ten seconds, is idle in the time) s
5 445 M
( interval [2, 3], and then restarts. At three seconds, a number of) s
5 434 M
( HTTP flows are started. The min, max and median of the) s
5 423 M
( request/response time of these HTTP flows would be plotted. Do) s
5 412 M
( the request/response times of these HTTP flows differ? If so, by) s
5 401 M
( how much?) s
5 379 M
( o High-congestion test: In a worst-case scenario with high) s
5 368 M
( congestion, all flows use TFRC, with a range of arrival times and) s
5 357 M
( idle times. The simulations are run both with and without Faster) s
5 346 M
( Restart. How does the use of Faster Restart affect the aggregate) s
5 335 M
( packet drop rate?) s
5 313 M
( o Transient changes: The first worst-case scenario with transient) s
5 302 M
( changes includes a routing change, where the new path has less) s
5 291 M
( bandwidth than the old path. The second scenario with transient) s
5 280 M
( changes includes transient congestion from a sudden increase in) s
5 269 M
( traffic. This increase in traffic could be from long-lived TCP) s
5 258 M
( traffic, or from higher-priority traffic, or from many new TFRC) s
5 247 M
( sessions. The transient congestion could be particularly severe) s
5 236 M
( if the congested link is an access link instead of a backbone) s
5 225 M
( link. The third scenario with transient changes could include a) s
5 214 M
( wireless link with variable bandwidth, as discussed earlier in) s
5 203 M
( Section 4. A fourth scenario would involve a mobility event that) s
5 192 M
( results in an increase in the round-trip time. In all cases, the) s
5 181 M
( simulations are run both with and without Faster Restart. How) s
5 170 M
( does the use of Faster Restart affect the aggregate packet drop) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 20]) s
_R
S
%%Page: (21) 21
%%BeginPageSetup
_S
24 24 translate
/pagenum 21 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
( rate?) s
5 665 M
( o An ideal scenario showing the benefits of Faster Restart: A) s
5 654 M
( scenario with an uncongested network, just a few TFRC flows,) s
5 643 M
( comparing the per-packet delay distribution with and without) s
5 632 M
( Faster Restart. Without Faster Restart, there should be a few) s
5 621 M
( packets in each flow with very large delay times, from waiting at) s
5 610 M
( the sender until they can be sent.) s
5 588 M
( o A scenario showing the benefits \(to the flow, not to competing) s
5 577 M
( traffic\) of padding during idle periods: Are there any scenarios) s
5 566 M
( where Faster Restart *increases* a flow's incentives to pad its) s
5 555 M
( sending rate during idle or under-utilized periods?) s
5 522 M
(Authors' Addresses) s
5 500 M
( Eddie Kohler) s
5 489 M
( 4531C Boelter Hall) s
5 478 M
( UCLA Computer Science Department) s
5 467 M
( Los Angeles, CA 90095) s
5 456 M
( USA) s
5 434 M
( Email: kohler@cs.ucla.edu) s
5 412 M
( Sally Floyd) s
5 401 M
( ICSI Center for Internet Research) s
5 390 M
( 1947 Center Street, Suite 600) s
5 379 M
( Berkeley, CA 94704) s
5 368 M
( USA) s
5 346 M
( Email: floyd@icir.org) s
5 324 M
( Arjuna Sathiaseelan) s
5 313 M
( Electronics Research Group) s
5 302 M
( University of Aberdeen) s
5 291 M
( Aberdeen) s
5 280 M
( UK) s
5 258 M
( Email: arjuna@erg.abdn.ac.uk) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 21]) s
_R
S
%%Page: (22) 22
%%BeginPageSetup
_S
24 24 translate
/pagenum 22 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(INTERNET-DRAFT Faster Restart for TFRC November 2007) s
5 687 M
(Full Copyright Statement) s
5 665 M
( Copyright \(C\) The IETF Trust \(2007\).) s
5 643 M
( This document is subject to the rights, licenses and restrictions) s
5 632 M
( contained in BCP 78, and except as set forth therein, the authors) s
5 621 M
( retain all their rights.) s
5 599 M
( This document and the information contained herein are provided on an) s
5 588 M
( "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS) s
5 577 M
( OR IS SPONSORED BY \(IF ANY\), THE INTERNET SOCIETY, THE IETF TRUST AND) s
5 566 M
( THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS) s
5 555 M
( OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF) s
5 544 M
( THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED) s
5 533 M
( WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.) s
5 511 M
(Intellectual Property) s
5 489 M
( The IETF takes no position regarding the validity or scope of any) s
5 478 M
( Intellectual Property Rights or other rights that might be claimed to) s
5 467 M
( pertain to the implementation or use of the technology described in) s
5 456 M
( this document or the extent to which any license under such rights) s
5 445 M
( might or might not be available; nor does it represent that it has) s
5 434 M
( made any independent effort to identify any such rights. Information) s
5 423 M
( on the procedures with respect to rights in RFC documents can be) s
5 412 M
( found in BCP 78 and BCP 79.) s
5 390 M
( Copies of IPR disclosures made to the IETF Secretariat and any) s
5 379 M
( assurances of licenses to be made available, or the result of an) s
5 368 M
( attempt made to obtain a general license or permission for the use of) s
5 357 M
( such proprietary rights by implementers or users of this) s
5 346 M
( specification can be obtained from the IETF on-line IPR repository at) s
5 335 M
( http://www.ietf.org/ipr.) s
5 313 M
( The IETF invites any interested party to bring to its attention any) s
5 302 M
( copyrights, patents or patent applications, or other proprietary) s
5 291 M
( rights that may cover technology that may be required to implement) s
5 280 M
( this standard. Please address the information to the IETF at ietf-) s
5 269 M
( ipr@ietf.org.) s
5 126 M
(Kohler, et al. Expires: May 2008 [Page 22]) s
_R
S
%%Page: (23) 23
%%BeginPageSetup
_S
24 24 translate
/pagenum 23 def
/fname (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/fdir () def
/ftail (draft-ietf-dccp-tfrc-faster-restart-05.txt) def
/user_header_p false def
%%EndPageSetup
_R
S
%%Trailer
%%Pages: 23
%%DocumentNeededResources: font Courier-Bold Courier
%%EOF
| PAFTECH AB 2003-2026 | 2026-04-22 23:34:44 |