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 2003-20262026-04-22 23:34:44