One document matched: draft-ietf-tsvwg-ecnsyn-00.ps


%!PS-Adobe-3.0
%%BoundingBox: 24 24 588 768
%%Title: Enscript Output
%%For: Sally Floyd
%%Creator: GNU enscript 1.6.1
%%CreationDate: Wed Nov 16 13:46:44 2005
%%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 (ecn-syn.txt) def
/fdir () def
/ftail (ecn-syn.txt) def
/user_header_p false def
%%EndPageSetup
5 731 M
(Internet Engineering Task Force                            A. Kuzmanovic) s
5 720 M
(INTERNET DRAFT                                   Northwestern University) s
5 709 M
(draft-ietf-tsvwg-ecnsyn-00.txt                                  S. Floyd) s
5 698 M
(                                                                    ICIR) s
5 687 M
(                                                       K.K. Ramakrishnan) s
5 676 M
(                                                                    AT&T) s
5 665 M
(                                                           October, 2005) s
5 621 M
(   Adding Explicit Congestion Notification \(ECN\) Capability to TCP's) s
5 610 M
(                            SYN/ACK Packets) s
5 566 M
(Status of this Memo) s
5 544 M
(   By submitting this Internet-Draft, each author represents that any) s
5 533 M
(   applicable patent or other IPR claims of which he or she is aware) s
5 522 M
(   have been or will be disclosed, and any of which he or she becomes) s
5 511 M
(   aware will be disclosed, in accordance with Section 6 of BCP 79.) s
5 489 M
(   Internet-Drafts are working documents of the Internet Engineering) s
5 478 M
(   Task Force \(IETF\), its areas, and its working groups.  Note that) s
5 467 M
(   other groups may also distribute working documents as Internet-) s
5 456 M
(   Drafts.) s
5 434 M
(   Internet-Drafts are draft documents valid for a maximum of six months) s
5 423 M
(   and may be updated, replaced, or obsoleted by other documents at any) s
5 412 M
(   time.  It is inappropriate to use Internet-Drafts as reference) s
5 401 M
(   material or to cite them other than as "work in progress.") s
5 379 M
(   The list of current Internet-Drafts can be accessed at) s
5 368 M
(   http://www.ietf.org/ietf/1id-abstracts.txt.) s
5 346 M
(   The list of Internet-Draft Shadow Directories can be accessed at) s
5 335 M
(   http://www.ietf.org/shadow.html.) s
5 313 M
(   This Internet-Draft will expire on April 2006.) s
5 291 M
(Copyright Notice) s
5 269 M
(   Copyright \(C\) The Internet Society \(2005\). All Rights Reserved.) s
5 247 M
(Abstract) s
5 225 M
(   This draft specifies a modification to RFC 3168 to allow TCP SYN/ACK) s
5 214 M
(   packets to be ECN-Capable.  For TCP, RFC 3168 only specified setting) s
5 170 M
(Kuzmanovic                   Standards Track                    [Page 1]) s
_R
S
%%Page: (2) 2
%%BeginPageSetup
_S
24 24 translate
/pagenum 2 def
/fname (ecn-syn.txt) def
/fdir () def
/ftail (ecn-syn.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(draft-ietf-tsvwg-ecnsyn-00.txt                              October 2005) s
5 687 M
(   an ECN-Capable codepoint on data packets, and not on SYN and SYN/ACK) s
5 676 M
(   packets.  However, because of the high cost to the TCP transfer of) s
5 665 M
(   having a SYN/ACK packet dropped, with the resulting retransmit) s
5 654 M
(   timeout, this document is specifying the use of ECN for the SYN/ACK) s
5 643 M
(   packet itself, when sent in response to a SYN packet with the two ECN) s
5 632 M
(   flags set in the TCP header, indicating a willingness to use ECN.) s
5 621 M
(   Setting TCP SYN/ACK packets as ECN-Capable can be of great benefit to) s
5 610 M
(   the TCP connection, avoiding the severe penalty of a retransmit) s
5 599 M
(   timeout for a connection that has not yet started placing a load on) s
5 588 M
(   the network.  The sender of the SYN/ACK packet must respond to an ECN) s
5 577 M
(   mark by reducing its initial congestion window from two, three, or) s
5 566 M
(   four segments to one segment, reducing the subsequent load from that) s
5 555 M
(   connection on the network.) s
5 533 M
(   NOTE TO RFC EDITOR: PLEASE DELETE THIS NOTE UPON PUBLICATION.) s
5 511 M
(   Changes from draft-kuzmanovic-ecn-syn-00.txt:) s
5 489 M
(   * Changed name of draft to draft-ietf-twvsg-ecnsyn.) s
5 467 M
(   END OF NOTE TO RFC EDITOR.) s
5 445 M
(1.  Conventions) s
5 423 M
(   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",) s
5 412 M
(   "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this) s
5 401 M
(   document are to be interpreted as described in [RFC 2119].) s
5 379 M
(1.  Introduction) s
5 357 M
(   TCP's congestion control mechanism has primarily used packet loss as) s
5 346 M
(   the congestion indication, with packets dropped when buffers) s
5 335 M
(   overflow.  With such tail-drop mechanisms, the packet delay can be) s
5 324 M
(   high, as the queue at bottleneck routers can be fairly large.) s
5 313 M
(   Dropping packets only when the queue overflows, and having TCP react) s
5 302 M
(   only to such losses, results in:) s
5 291 M
(   1\) significantly higher packet delay;) s
5 280 M
(   2\) unnecessarily many packet losses; and) s
5 269 M
(   3\) unfairness due to synchronization effects.) s
5 247 M
(   The adoption of Active Queue Management \(AQM\) mechanisms allows) s
5 236 M
(   better control of bottleneck queues.  This use of AQM has the) s
5 225 M
(   following potential benefits:) s
5 214 M
(   1\) better control of the queue, with reduced queueing delay;) s
5 203 M
(   2\) fewer packet drops; and) s
5 192 M
(   3\) better fairness because of fewer synchronization effects.) s
5 170 M
(   With the adoption of ECN, performance may be further improved.  When) s
5 126 M
(Kuzmanovic                   Standards Track                    [Page 2]) s
_R
S
%%Page: (3) 3
%%BeginPageSetup
_S
24 24 translate
/pagenum 3 def
/fname (ecn-syn.txt) def
/fdir () def
/ftail (ecn-syn.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(draft-ietf-tsvwg-ecnsyn-00.txt                              October 2005) s
5 687 M
(   the router detects congestion before buffer overflow, the router can) s
5 676 M
(   provide a congestion indication either by dropping a packet, or by) s
5 665 M
(   setting the Congestion Experienced \(CE\) codepoint in the  Explicit) s
5 654 M
(   Congestion Notification \(ECN\) field in the IP header [RFC3168].  The) s
5 643 M
(   IETF has standardized the use of the Congestion Experienced \(CE\)) s
5 632 M
(   codepoint in the IP header for routers to indicate congestion.  For) s
5 621 M
(   incremental deployment and backwards compatibility, the RFC on ECN) s
5 610 M
(   [RFC 3168] specifies that routers may mark ECN-capable packets that) s
5 599 M
(   would otherwise have been dropped, using the Congestion Experienced) s
5 588 M
(   codepoint in the ECN field.  The use of ECN allows TCP to react to) s
5 577 M
(   congestion while avoiding unnecessary retransmissions and, in some) s
5 566 M
(   cases, unnecessary retransmit timeouts.  Thus, using ECN has several) s
5 555 M
(   benefits:) s
5 533 M
(   1\) For short transfers, a TCP connection's congestion window may be) s
5 522 M
(   small.  For example, if the current window contains only one packet,) s
5 511 M
(   and that packet is dropped, TCP will have to wait for a retransmit) s
5 500 M
(   timeout to recover, reducing its overall throughput.  Similarly, if) s
5 489 M
(   the current window contains only a few packets and one of those) s
5 478 M
(   packets is dropped, there might not be enough duplicate) s
5 467 M
(   acknowledgements for a fast retransmission, and the sender might have) s
5 456 M
(   to wait for a delay of several round-trip times using Limited) s
5 445 M
(   Transmit [RFC3042].  With the use of ECN, short flows are less likely) s
5 434 M
(   to have packets dropped, sometimes avoiding unnecessary delays or) s
5 423 M
(   costly retransmit timeouts.) s
5 401 M
(   2\) While longer flows may not see substantially improved throughput) s
5 390 M
(   with the use of ECN, they experience lower loss. This may benefit TCP) s
5 379 M
(   applications that are latency- and loss-sensitive, because of the) s
5 368 M
(   avoidance of retransmissions.) s
5 346 M
(   RFC 3168 only specified marking the Congestion Experienced codepoint) s
5 335 M
(   on TCP's data packets, and not on SYN and SYN/ACK packets.  RFC 3168) s
5 324 M
(   specified the negotiation of the use of ECN between the two TCP end-) s
5 313 M
(   points in the TCP SYN and SYN-ACK exchange, using flags in the TCP) s
5 302 M
(   header.  Erring on the side of being conservative, RFC 3168 did not) s
5 291 M
(   specify the use of ECN for the SYN/ACK packet itself.  However,) s
5 280 M
(   because of the high cost to the TCP transfer of having a SYN/ACK) s
5 269 M
(   packet dropped, with the resulting retransmit timeout, this document) s
5 258 M
(   is specifying the use of ECN for the SYN/ACK packet itself.  This can) s
5 247 M
(   be of great benefit to the TCP connection, avoiding the severe) s
5 236 M
(   penalty of a retransmit timeout for a connection that has not yet) s
5 225 M
(   started placing a load on the network.  The sender of the SYN/ACK) s
5 214 M
(   packet must respond to an ECN mark by reducing its initial congestion) s
5 203 M
(   window from two, three, or four segments to one segment, reducing the) s
5 192 M
(   subsequent load from that connection on the network.) s
5 170 M
(   The use of ECN for SYN/ACK packets has the following potential) s
5 126 M
(Kuzmanovic                   Standards Track                    [Page 3]) s
_R
S
%%Page: (4) 4
%%BeginPageSetup
_S
24 24 translate
/pagenum 4 def
/fname (ecn-syn.txt) def
/fdir () def
/ftail (ecn-syn.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(draft-ietf-tsvwg-ecnsyn-00.txt                              October 2005) s
5 687 M
(   benefits:) s
5 676 M
(   1\) Avoidance of a retransmit timeout;) s
5 665 M
(   2\) Improvement in the throughput of short connections.) s
5 643 M
(   This draft specifies a modification to RFC 3168 to allow TCP SYN/ACK) s
5 632 M
(   packets to be ECN-Capable.  Section 2 contains the specification of) s
5 621 M
(   the change, while Section 3 discusses some of the issues, and Section) s
5 610 M
(   4 discusses related work.  Section 5 contains an evaluation of the) s
5 599 M
(   proposed change.) s
5 577 M
(2.  Proposal) s
5 555 M
(   This section specifies the modification to RFC 3168 to allow TCP) s
5 544 M
(   SYN/ACK packets to be ECN-Capable.  We use the following terminology) s
5 533 M
(   from RFC 3168:) s
5 511 M
(   The ECN field in the IP header:) s
5 500 M
(   o  CE: the Congestion Experienced codepoint; and) s
5 489 M
(   o  ECT: either one of the two ECN-Capable Transport codepoints.) s
5 467 M
(   The ECN flags in the TCP header:) s
5 456 M
(   o  CWR: the Congestion Window Reduced flag; and) s
5 445 M
(   o  ECE: the ECN-Echo flag.) s
5 423 M
(   ECN-setup packets:) s
5 412 M
(   o  ECN-setup SYN packet: a SYN packet with the ECE and CWR flags;) s
5 401 M
(   o  ECN-setup SYN-ACK packet: a SYN-ACK packet with ECE but not CWR.) s
5 379 M
(   RFC 3168 in Section 6.1.1. states that "A host MUST NOT set ECT on) s
5 368 M
(   SYN or SYN-ACK packets." In this section, we specify that a TCP node) s
5 357 M
(   MAY respond to an ECN-setup SYN packet by setting ECT in the) s
5 346 M
(   responding ECN-setup SYN/ACK packet, indicating to routers that the) s
5 335 M
(   SYN/ACK packet is ECN-Capable.  This allows a congested router along) s
5 324 M
(   the path to mark the packet instead of dropping the packet as an) s
5 313 M
(   indication of congestion.) s
5 291 M
(   Assume that TCP node A transmits to TCP node B an ECN-setup SYN) s
5 280 M
(   packet, indicating willingness to use ECN for this connection.  As) s
5 269 M
(   specified by RFC 3168, if TCP node B is willing to use ECN, node B) s
5 258 M
(   responds with an ECN-setup SYN-ACK packet.) s
5 236 M
(   Table 1 shows an interchange with the SYN/ACK packet dropped by a) s
5 225 M
(   congested router.  Node B waits for a retransmit timeout, and then) s
5 214 M
(   retransmits the SYN/ACK packet.) s
5 126 M
(Kuzmanovic                   Standards Track                    [Page 4]) s
_R
S
%%Page: (5) 5
%%BeginPageSetup
_S
24 24 translate
/pagenum 5 def
/fname (ecn-syn.txt) def
/fdir () def
/ftail (ecn-syn.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(draft-ietf-tsvwg-ecnsyn-00.txt                              October 2005) s
5 687 M
(     ---------------------------------------------------------------) s
5 676 M
(        TCP Node A             Router                  TCP Node B) s
5 665 M
(        ----------             ------                  ----------) s
5 643 M
(        ECN-setup SYN packet --->) s
5 632 M
(                                         ECN-setup SYN packet --->) s
5 610 M
(                              <--- ECN-setup SYN/ACK, possibly ECT) s
5 599 M
(                                                3-second timer set) s
5 588 M
(                            SYN/ACK dropped               .) s
5 577 M
(                                                          .) s
5 566 M
(                                                          .) s
5 555 M
(                                            3-second timer expires) s
5 544 M
(                                   <--- ECN-setup SYN/ACK, not ECT) s
5 533 M
(        <--- ECN-setup SYN/ACK) s
5 522 M
(        Data/ACK --->) s
5 511 M
(                                                     Data/ACK --->) s
5 500 M
(                                  <--- Data \(one to four segments\)) s
5 489 M
(     ---------------------------------------------------------------) s
5 467 M
(        Table 1: SYN exchange with the SYN/ACK packet dropped.) s
5 434 M
(   If the SYN/ACK packet is dropped in the network, the TCP host \(node) s
5 423 M
(   B\) responds by waiting three seconds for the retransmit timer to) s
5 412 M
(   expire [RFC2988].  If a SYN/ACK packet with the ECT codepoint is) s
5 401 M
(   dropped, the TCP node SHOULD resend the SYN/ACK packet without the) s
5 390 M
(   ECN-Capable codepoint.  \(Although we are not aware of any middleboxes) s
5 379 M
(   that drop SYN/ACK packets that contain an ECN-Capable codepoint in) s
5 368 M
(   the IP header, we have learned to design our protocols defensively in) s
5 357 M
(   this regard [RFC3360].\)) s
5 335 M
(   Table 2 shows an interchange with the SYN/ACK packet sent as ECN-) s
5 324 M
(   Capable, and ECN-marked instead of dropped at the congested router.) s
5 126 M
(Kuzmanovic                   Standards Track                    [Page 5]) s
_R
S
%%Page: (6) 6
%%BeginPageSetup
_S
24 24 translate
/pagenum 6 def
/fname (ecn-syn.txt) def
/fdir () def
/ftail (ecn-syn.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(draft-ietf-tsvwg-ecnsyn-00.txt                              October 2005) s
5 687 M
(     ---------------------------------------------------------------) s
5 676 M
(        TCP Node A             Router                  TCP Node B) s
5 665 M
(        ----------             ------                  ----------) s
5 643 M
(        ECN-setup SYN packet --->) s
5 632 M
(                                        ECN-setup SYN packet --->) s
5 610 M
(                                      <--- ECN-setup SYN/ACK, ECT) s
5 599 M
(                           <--- Sets CE on SYN/ACK) s
5 588 M
(        <--- ECN-setup SYN/ACK, CE) s
5 566 M
(        Data/ACK, ECN-Echo --->) s
5 555 M
(                                          Data/ACK, ECN-Echo --->) s
5 544 M
(                                   Window reduced to one segment.) s
5 533 M
(                                <--- Data, CWR \(one segment only\)) s
5 522 M
(     ---------------------------------------------------------------) s
5 500 M
(        Table 2: SYN exchange with the SYN/ACK packet marked.) s
5 467 M
(   If the receiving node \(node A\) receives a SYN/ACK packet that has) s
5 456 M
(   been marked by the congested router, with the CE codepoint set, the) s
5 445 M
(   receiving node MUST respond by setting the ECN-Echo flag in the TCP) s
5 434 M
(   header of the responding ACK packet.  As specified in RFC 3168, the) s
5 423 M
(   receiving node continues to set the ECN-Echo flag in packets until it) s
5 412 M
(   receives a packet with the CWR flag set.) s
5 390 M
(   When the sending node \(node B\) receives the ECN-Echo packet reporting) s
5 379 M
(   the Congestion Experienced indication in the SYN/ACK packet, the node) s
5 368 M
(   MUST set the initial congestion window to one segment, instead of two) s
5 357 M
(   segments as allowed by [RFC2414], or three or four segments allowed) s
5 346 M
(   by [RFC3390].  If the sending node \(node B\) was going to use an) s
5 335 M
(   initial window of one segment, and receives an ECN-Echo packet) s
5 324 M
(   informing it of a Congestion Experienced indication on its SYN/ACK) s
5 313 M
(   packet, the sending node MAY continue to send with an initial window) s
5 302 M
(   of one segment, without waiting for a retransmit timeout.  We note) s
5 291 M
(   that this updates RFC 3168, which specifies that "the sending TCP) s
5 280 M
(   MUST reset the retransmit timer on receiving the ECN-Echo packet when) s
5 269 M
(   the congestion window is one."  As specified by RFC 3168, the sending) s
5 258 M
(   node \(node B\) also sets the CWR flag in the TCP header of the next) s
5 247 M
(   packet sent, to acknowledge its receipt of and reaction to the ECN-) s
5 236 M
(   Echo flag.) s
5 214 M
(3.  Discussion) s
5 192 M
(   Motivation:) s
5 181 M
(   The rationale for the proposed change is the following.  When node B) s
5 170 M
(   receives a TCP SYN packet with ECN-Echo bit set in the TCP header,) s
5 126 M
(Kuzmanovic                   Standards Track                    [Page 6]) s
_R
S
%%Page: (7) 7
%%BeginPageSetup
_S
24 24 translate
/pagenum 7 def
/fname (ecn-syn.txt) def
/fdir () def
/ftail (ecn-syn.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(draft-ietf-tsvwg-ecnsyn-00.txt                              October 2005) s
5 687 M
(   this indicates that node A is ECN-capable. If node B is also ECN-) s
5 676 M
(   capable, there are no obstacles to immediately setting one of the) s
5 665 M
(   ECN-Capable codepoints in the IP header in the responding TCP SYN/ACK) s
5 654 M
(   packet.) s
5 632 M
(   There can be a great benefit in setting an ECN-capable codepoint in) s
5 621 M
(   SYN/ACK packets, as is discussed further in Section 4.  Congestion is) s
5 610 M
(   most likely to occur in the server-to-client direction.  As a result,) s
5 599 M
(   setting an ECN-capable codepoint in SYN/ACK packets can reduce the) s
5 588 M
(   occurence of three-second retransmit timeouts resulting from the drop) s
5 577 M
(   of SYN/ACK packets.) s
5 555 M
(   Flooding attacks:) s
5 544 M
(   Setting an ECN-Capable codepoint in the responding TCP SYN/ACK) s
5 533 M
(   packets does not raise any novel security vulnerabilities.  For) s
5 522 M
(   example, provoking servers or hosts to send SYN/ACK packets to a) s
5 511 M
(   third party in order to perform a "SYN/ACK flood" attack would be) s
5 500 M
(   greatly inefficient.  Third parties would immediately drop such) s
5 489 M
(   packets, since they would know that they didn't generate the TCP SYN) s
5 478 M
(   packets in the first place.  Moreover, such SYN/ACK attacks would) s
5 467 M
(   have the same signatures as the existing TCP SYN attacks. Provoking) s
5 456 M
(   servers or hosts to reply with SYN/ACK packets in order to congest a) s
5 445 M
(   certain link would also be highly inefficient because SYN ACK packets) s
5 434 M
(   are small in size.) s
5 412 M
(   The TCP SYN packet:) s
5 401 M
(   There are several reasons why an ECN-Capable codepoint MUST NOT be) s
5 390 M
(   set in the IP header of the initiating TCP SYN packet.  First, when) s
5 379 M
(   the TCP SYN packet is sent, there are no guarantees that the other) s
5 368 M
(   TCP endpoint \(node B in Table 2\) is ECN-capable, or that it would be) s
5 357 M
(   able to understand and react if the ECN CE codepoint was set by a) s
5 346 M
(   congested router.) s
5 324 M
(   Second, the ECN-Capable codepoint in TCP SYN packets could be misused) s
5 313 M
(   by malicious clients to `improve' the well-known TCP SYN attack. By) s
5 302 M
(   setting an ECN-Capable codepoint in TCP SYN packets, a malicious host) s
5 291 M
(   might be able to inject a large number of TCP SYN packets through a) s
5 280 M
(   potentially congested ECN-enabled router, congesting it even further.) s
5 258 M
(   For both these reasons, we continue the restriction that the TCP SYN) s
5 247 M
(   packet MUST NOT have the ECN-Capable codepoint in the IP header set.) s
5 225 M
(   Backwards compatibility:) s
5 214 M
(   If there are some older TCP implementations that don't respond to the) s
5 203 M
(   Congestion Experienced codepoint in a SYN/ACK packet, that would not) s
5 192 M
(   be an insurmountable problem.  It would mean that the sender of the) s
5 181 M
(   SYN/ACK packet would not reduce the initial congestion window from) s
5 170 M
(   two, three, or four segments down to one segment, as it should.) s
5 126 M
(Kuzmanovic                   Standards Track                    [Page 7]) s
_R
S
%%Page: (8) 8
%%BeginPageSetup
_S
24 24 translate
/pagenum 8 def
/fname (ecn-syn.txt) def
/fdir () def
/ftail (ecn-syn.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(draft-ietf-tsvwg-ecnsyn-00.txt                              October 2005) s
5 687 M
(   However, the TCP sender would still respond correctly to any) s
5 676 M
(   subsequent CE indications on data packets later on in the connection.) s
5 654 M
(   SYN/ACK packets and packet size:) s
5 643 M
(   There are a number of router buffer architectures that have smaller) s
5 632 M
(   dropping rates for small \(SYN\) packets than for large \(data\) packets.) s
5 621 M
(   For example, for a Drop Tail queue in units of packets, where each) s
5 610 M
(   packet takes a single slot in the buffer regardless of packet size,) s
5 599 M
(   small and large packets are equally likely to be dropped.  However,) s
5 588 M
(   for a Drop Tail queue in units of bytes, small packets are less) s
5 577 M
(   likely to be dropped than are large ones.  Similarly, for RED in) s
5 566 M
(   packet mode, small and large packets are equally likely to be dropped) s
5 555 M
(   or marked, while for RED in byte mode, a packet's chance of being) s
5 544 M
(   dropped or marked is proportional to the packet size in bytes.) s
5 522 M
(   For a congested router with an AQM mechanism in byte mode, where a) s
5 511 M
(   packet's chance of being dropped or marked is proportional to the) s
5 500 M
(   packet size in bytes, the drop or marking rate for TCP SYN/ACK) s
5 489 M
(   packets should generally be low.  In this case, the benefit of making) s
5 478 M
(   SYN/ACK packets ECN-Capable should be similarly moderate.  However,) s
5 467 M
(   for a congested router with a Drop Tail queue in units of packets or) s
5 456 M
(   with an AQM mechanism in packet mode, and with no priority queueing) s
5 445 M
(   for smaller packets, small and large packets should have the same) s
5 434 M
(   probability of being dropped or marked.  In such a case, making) s
5 423 M
(   SYN/ACK packets ECN-Capable should be of significant benefit.) s
5 401 M
(   We believe that there are a wide range of behaviors in the real world) s
5 390 M
(   in terms of the drop or mark behavior at routers as a function of) s
5 379 M
(   packet size [Tools, Section 10].  We note that all of these) s
5 368 M
(   alternatives listed above are available in the NS simulator \(Drop) s
5 357 M
(   Tail queues are by default in units of packets, while the default for) s
5 346 M
(   RED queue management has been changed from packet mode to byte mode\).) s
5 324 M
(4.  Related Work) s
5 302 M
(   The addition of ECN-capability to TCP's SYN/ACK packets was proposed) s
5 291 M
(   in [ECN+].  The paper includes an extensive set of simulation and) s
5 280 M
(   testbed experiments to evaluate the effects of the proposal, using) s
5 269 M
(   several Active Queue Management \(AQM\) mechanisms, including Random) s
5 258 M
(   Early Detection \(RED\) [RED], Random Exponential Marking \(REM\) [REM],) s
5 247 M
(   and Proportional Integrator \(PI\) [PI].  The performance measures were) s
5 236 M
(   the end-to-end response times for each request/response pair, and the) s
5 225 M
(   aggregate throughput on the bottleneck link.  The end-to-end response) s
5 214 M
(   time was computed as the time from the moment when the request for) s
5 203 M
(   the file is sent to the server, until that file is successfully) s
5 192 M
(   downloaded by the client.) s
5 170 M
(   The measurements from [ECN+] showed that setting an ECN-Capable) s
5 126 M
(Kuzmanovic                   Standards Track                    [Page 8]) s
_R
S
%%Page: (9) 9
%%BeginPageSetup
_S
24 24 translate
/pagenum 9 def
/fname (ecn-syn.txt) def
/fdir () def
/ftail (ecn-syn.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(draft-ietf-tsvwg-ecnsyn-00.txt                              October 2005) s
5 687 M
(   codepoint in the IP packet header in TCP SYN/ACK packets) s
5 676 M
(   systematically improves performance with all evaluated AQM schemes.) s
5 665 M
(   When SYN/ACK packets at a congested router are ECN-marked instead of) s
5 654 M
(   dropped, this can avoid a long initial retransmit timeout, improving) s
5 643 M
(   the response time for the affected flow dramatically.) s
5 621 M
(   [ECN+] showed that the impact on aggregate throughput can also be) s
5 610 M
(   quite significant, because marking SYN ACK packets can prevent larger) s
5 599 M
(   flows from suffering long timeouts before being "admitted" into the) s
5 588 M
(   network.  In addition, the testbed measurements from [ECN+] showed) s
5 577 M
(   that Web servers setting the ECN-Capable codepoint in TCP SYN/ACK) s
5 566 M
(   packets could serve more requests.) s
5 544 M
(   As a final step, [ECN+] explored the co-existence of flows that do) s
5 533 M
(   and don't set the ECN-capable codepoint in TCP SYN/ACK packets.  The) s
5 522 M
(   results in [ECN+] confirmed that both types of flows can coexist;) s
5 511 M
(   flows that apply the change improve their end-to-end performance,) s
5 500 M
(   while the performance degradation for flows that don't apply the) s
5 489 M
(   change, as a result of the flows that do apply the change, is) s
5 478 M
(   marginal.) s
5 456 M
(5.  Evaluation) s
5 434 M
(   The addition of ECN-capability to SYN/ACK packets could be of) s
5 423 M
(   significant benefit for those ECN connections that would have had the) s
5 412 M
(   SYN/ACK packet dropped in the network, and for which the ECN-) s
5 401 M
(   Capability would allow the SYN/ACK to be marked rather than dropped.) s
5 379 M
(   The percent of SYN/ACK packets on a link can be quite high. In) s
5 368 M
(   particular, measurements on links dominated by Web traffic indicate) s
5 357 M
(   that 15-20% of the packets can be SYN/ACK packets [SCJO01].) s
5 335 M
(   The benefit of adding ECN-capability to SYN/ACK packets depends in) s
5 324 M
(   part on the size of the data transfer.  The drop of a SYN/ACK packet) s
5 313 M
(   can increase the download time of a short file by an order of) s
5 302 M
(   magnitude, by requiring a three-second retransmit timeout.  For) s
5 291 M
(   longer-lived flows, the effect of a dropped SYN/ACK packet on file) s
5 280 M
(   download time is less dramatic.  However, even for longer-lived) s
5 269 M
(   flows, the addition of ECN-capability to SYN/ACK packets can improve) s
5 258 M
(   the fairness among long-lived flows, as newly-arriving flows would be) s
5 247 M
(   less likely to have to wait for retransmit timeouts.) s
5 225 M
(   The question that arises of course is what fraction of connections) s
5 214 M
(   would see the benefit from making SYN/ACK packets ECN-capable, in a) s
5 203 M
(   particular scenario?  Specifically:) s
5 181 M
(   \(1\) What fraction of arriving SYN/ACK packets are dropped at the) s
5 170 M
(   congested router when the SYN/ACK packets are not ECN-capable?) s
5 126 M
(Kuzmanovic                   Standards Track                    [Page 9]) s
_R
S
%%Page: (10) 10
%%BeginPageSetup
_S
24 24 translate
/pagenum 10 def
/fname (ecn-syn.txt) def
/fdir () def
/ftail (ecn-syn.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(draft-ietf-tsvwg-ecnsyn-00.txt                              October 2005) s
5 687 M
(   \(2\) Of those SYN/ACK packets that are dropped, what fraction of those) s
5 676 M
(   drops would have been ECN-marks instead of drops if the SYN/ACK) s
5 665 M
(   packets had been ECN-capable?) s
5 643 M
(   To answer \(1\), it is necessary to consider not only the level of) s
5 632 M
(   congestion but also the queue architecture at the congested link.  As) s
5 621 M
(   described in Section 3 above, for some queue architectures small) s
5 610 M
(   packets are less likely to be dropped than large ones.  In such an) s
5 599 M
(   environment, SYN/ACK packets would have lower packet drop rates;) s
5 588 M
(   question \(1\) could not necessarily be inferred from the overall) s
5 577 M
(   packet drop rate, but could be answered by measuring the drop rate) s
5 566 M
(   for SYN/ACK packets directly.  In such an environment, adding ECN-) s
5 555 M
(   capability to SYN/ACK packets would be of less dramatic benefit than) s
5 544 M
(   in environments where all packets are equally likely to be dropped) s
5 533 M
(   regardless of packet size.) s
5 511 M
(   As question \(2\) implies, even if all of the SYN/ACK packets were ECN-) s
5 500 M
(   capable, there could still be some SYN/ACK packets dropped instead of) s
5 489 M
(   marked at the congested link; the full answer to question \(2\) depends) s
5 478 M
(   on the details of the queue management mechanism at the router.  If) s
5 467 M
(   congestion is sufficiently bad, and the queue management mechanism) s
5 456 M
(   cannot prevent the buffer from overflowing, then SYN/ACK packets will) s
5 445 M
(   be dropped rather than marked upon buffer overflow whether or not) s
5 434 M
(   they are ECN-capable.) s
5 412 M
(   For some AQM mechanisms, ECN-capable packets are marked instead of) s
5 401 M
(   dropped any time this is possible, that is, any time the buffer is) s
5 390 M
(   not yet full.  For other AQM mechanisms however, such as the RED) s
5 379 M
(   mechanism as recommended in [RED], packets are dropped rather than) s
5 368 M
(   marked when the packet drop/mark rate exceeds a certain threshold,) s
5 357 M
(   e.g., 10%, even if the packets are ECN-capable.  For a router with) s
5 346 M
(   such an AQM mechanism, when congestion is sufficiently severe to) s
5 335 M
(   cause a high drop/mark rate, some SYN/ACK packets would be dropped) s
5 324 M
(   instead of marked whether or not they were ECN-capable.) s
5 302 M
(   Thus, the degree of benefit of adding ECN-Capability to SYN/ACK) s
5 291 M
(   packets depends not only on the overall packet drop rate in the) s
5 280 M
(   network, but also on the queue management architecture at the) s
5 269 M
(   congested link.) s
5 247 M
(6.  Security Considerations) s
5 225 M
(   TCP packets carrying the ECT codepoint in IP headers can be marked) s
5 214 M
(   rather than dropped by ECN-capable routers. This raises several) s
5 203 M
(   security concerns that we discuss below.) s
5 181 M
(   TCP SYN flooding attacks:) s
5 170 M
(   By setting an ECN-Capable codepoint in TCP SYN packets, a malicious) s
5 126 M
(Kuzmanovic                   Standards Track                   [Page 10]) s
_R
S
%%Page: (11) 11
%%BeginPageSetup
_S
24 24 translate
/pagenum 11 def
/fname (ecn-syn.txt) def
/fdir () def
/ftail (ecn-syn.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(draft-ietf-tsvwg-ecnsyn-00.txt                              October 2005) s
5 687 M
(   host might be able to inject a large number of TCP SYN packets) s
5 676 M
(   through a potentially congested ECN-enabled router, congesting it) s
5 665 M
(   even further. This is one of the reasons why an ECN-Capable codepoint) s
5 654 M
(   MUST NOT be set in the IP header of the initiating TCP SYN packet.) s
5 643 M
(   On the other hand, as discussed in Section 3 above, setting an ECN-) s
5 632 M
(   Capable codepoint in the responding TCP SYN/ACK packet does not raise) s
5 621 M
(   any novel security vulnerabilities.) s
5 599 M
(   "Bad" middleboxes:) s
5 588 M
(   While there is no evidence that any middleboxes drop SYN/ACK packets) s
5 577 M
(   that contain an ECN-Capable codepoint in the IP header, such behavior) s
5 566 M
(   cannot be excluded [RFC3360]. Thus, if a SYN/ACK packet with the ECT) s
5 555 M
(   codepoint is dropped, the TCP node SHOULD resend the SYN/ACK packet) s
5 544 M
(   without the ECN-Capable codepoint.) s
5 522 M
(   Congestion collapse:) s
5 511 M
(   Because TCP SYN/ACK packets carrying an ECT codepoint could be ECN-) s
5 500 M
(   marked instead of dropped at an ECN-capable router, the concern is) s
5 489 M
(   whether this can either invoke congestion, or worsen performance in) s
5 478 M
(   highly congested scenarios.  This is not a problem because after) s
5 467 M
(   learning that the SYN/ACK packet was ECN-marked, the sender of that) s
5 456 M
(   packet will only send one data packet; in the case that this data) s
5 445 M
(   packet is ECN-marked, the sender will wait for a retransmission) s
5 434 M
(   timeout.  In addition, routers are free to drop rather than mark) s
5 423 M
(   arriving packets in times of high congestion, regardless of whether) s
5 412 M
(   the packets are ECN-capable.) s
5 390 M
(7.  Conclusions) s
5 368 M
(   This draft specifies a modification to RFC 3168 to allow TCP nodes to) s
5 357 M
(   send SYN/ACK packets as being ECN-Capable.  Making the SYN/ACK packet) s
5 346 M
(   ECN-Capable avoids the high cost to a TCP transfer when a SYN/ACK) s
5 335 M
(   packet is dropped by a congested router, by avoiding the resulting) s
5 324 M
(   retransmit timeout.  This improves the throughput of short) s
5 313 M
(   connections.  The sender of the SYN/ACK packet responds to an ECN) s
5 302 M
(   mark by reducing its initial congestion window from two, three, or) s
5 291 M
(   four segments to one segment, reducing the subsequent load from that) s
5 280 M
(   connection on the network.  The addition of ECN-capability to SYN/ACK) s
5 269 M
(   packets is particularly beneficial in the server-to-client direction,) s
5 258 M
(   where congestion is more likely to occur.  In this case, the initial) s
5 247 M
(   information provided by the ECN marking in the SYN/ACK packet enables) s
5 236 M
(   the server to more appropriately adjust the initial load it places on) s
5 225 M
(   the network.) s
5 126 M
(Kuzmanovic                   Standards Track                   [Page 11]) s
_R
S
%%Page: (12) 12
%%BeginPageSetup
_S
24 24 translate
/pagenum 12 def
/fname (ecn-syn.txt) def
/fdir () def
/ftail (ecn-syn.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(draft-ietf-tsvwg-ecnsyn-00.txt                              October 2005) s
5 687 M
(8.  Acknowledgements) s
5 665 M
(9.  Normative References) s
5 643 M
(   [RFC2414] M. Allman, S. Floyd, and C. Partridge, Increasing TCP's) s
5 632 M
(   Initial Window, RFC 2414, September 1998.) s
5 610 M
(   [RFC3168] K.K. Ramakrishnan, S. Floyd, and D. Black, The Addition of) s
5 599 M
(   Explicit Congestion Notification \(ECN\) to IP, RFC 3168, Proposed) s
5 588 M
(   Standard, September 2001.) s
5 566 M
(   [RFC3390] M. Allman, S. Floyd, and C. Partridge, Increasing TCP's) s
5 555 M
(   Initial Window, RFC 3390, October 2002.) s
5 533 M
(10.  Informative References) s
5 511 M
(   [ECN+] A. Kuzmanovic, The Power of Explicit Congestion Notification,) s
5 500 M
(   SIGCOMM 2005.) s
5 478 M
(   [PI] C. Hollot, V. Misra, W. Gong, and D. Towsley, On Designing) s
5 467 M
(   Improved Controllers for AQM Routers Supporting TCP Flows, INFOCOM,) s
5 456 M
(   June 2001.) s
5 434 M
(   [RED] S. Floyd and V. Jacobson, Random Early Detection Gateways for) s
5 423 M
(   Congestion Avoidance, IEEE/ACM Transactions on Networking, V.1, N.4,) s
5 412 M
(   1993.) s
5 390 M
(   [REM] S. Athuraliya, V. Li, S. Low, and Q Yin, REM: Active Queue) s
5 379 M
(   Management, IEEE Network, V.15, N. 3, May 2001.) s
5 357 M
(   [RFC2988] V. Paxson and M. Allman, Computing TCP's Retransmission) s
5 346 M
(   Timer, RFC 2988, November 2000.) s
5 324 M
(   [RFC3042] M. Allman, H. Balakrishnan, and S. Floyd, Enhancing TCP's) s
5 313 M
(   Loss Recovery Using Limited Transmit, RFC 3042, Proposed Standard,) s
5 302 M
(   January 2001.) s
5 280 M
(   [RFC3360] S. Floyd, Inappropriate TCP Resets Considered Harmful, RFC) s
5 269 M
(   3360, August 2002.) s
5 247 M
(   [SCJO01] F. Smith, F. Campos, K. Jeffay, D. Ott, What {TCP/IP}) s
5 236 M
(   Protocol Headers Can Tell us about the Web, SIGMETRICS, June 2001.) s
5 214 M
(   [Tools] S. Floyd and E. Kohler, Tools for the Evaluation of) s
5 203 M
(   Simulation and Testbed Scenarios, Internet-draft draft-irtf-tmrg-) s
5 192 M
(   tools-00, work in progress, September 2005.) s
5 126 M
(Kuzmanovic                   Standards Track                   [Page 12]) s
_R
S
%%Page: (13) 13
%%BeginPageSetup
_S
24 24 translate
/pagenum 13 def
/fname (ecn-syn.txt) def
/fdir () def
/ftail (ecn-syn.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(draft-ietf-tsvwg-ecnsyn-00.txt                              October 2005) s
5 687 M
(11.  IANA Considerations) s
5 665 M
(   There are no IANA considerations regarding this document.) s
5 643 M
(   AUTHORS' ADDRESSES) s
5 610 M
(      Aleksandar Kuzmanovic) s
5 599 M
(      Phone: +1 \(847\) 467-5519) s
5 588 M
(      Northwestern University) s
5 577 M
(      Email: akuzma@northwestern.edu) s
5 566 M
(      URL: http://cs.northwestern.edu/~a) s
5 544 M
(      Sally Floyd) s
5 533 M
(      Phone: +1 \(510\) 666-2989) s
5 522 M
(      ICIR \(ICSI Center for Internet Research\)) s
5 511 M
(      Email: floyd@icir.org) s
5 500 M
(      URL: http://www.icir.org/floyd/) s
5 478 M
(      K. K. Ramakrishnan) s
5 467 M
(      Phone: +1 \(973\) 360-8764) s
5 456 M
(      AT&T Labs Research) s
5 445 M
(      Email: kkrama@research.att.com) s
5 434 M
(      URL: http://www.research.att.com/info/kkrama) s
5 412 M
(Full Copyright Statement) s
5 390 M
(   Copyright \(C\) The Internet Society 2005.  This document is subject) s
5 379 M
(   to the rights, licenses and restrictions contained in BCP 78, and) s
5 368 M
(   except as set forth therein, the authors retain all their rights.) s
5 346 M
(   This document and the information contained herein are provided on) s
5 335 M
(   an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE) s
5 324 M
(   REPRESENTS OR IS SPONSORED BY \(IF ANY\), THE INTERNET SOCIETY AND THE) s
5 313 M
(   INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR) s
5 302 M
(   IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF) s
5 291 M
(   THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED) s
5 280 M
(   WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.) s
5 258 M
(Intellectual Property) s
5 236 M
(   The IETF takes no position regarding the validity or scope of any) s
5 225 M
(   Intellectual Property Rights or other rights that might be claimed) s
5 214 M
(   to pertain to the implementation or use of the technology described) s
5 203 M
(   in this document or the extent to which any license under such) s
5 192 M
(   rights might or might not be available; nor does it represent that) s
5 181 M
(   it has made any independent effort to identify any such rights.) s
5 170 M
(   Information on the procedures with respect to rights in RFC) s
5 126 M
(Kuzmanovic                   Standards Track                   [Page 13]) s
_R
S
%%Page: (14) 14
%%BeginPageSetup
_S
24 24 translate
/pagenum 14 def
/fname (ecn-syn.txt) def
/fdir () def
/ftail (ecn-syn.txt) def
/user_header_p false def
%%EndPageSetup
5 720 M
(draft-ietf-tsvwg-ecnsyn-00.txt                              October 2005) s
5 687 M
(   documents can be found in BCP 78 and BCP 79.) s
5 665 M
(   Copies of IPR disclosures made to the IETF Secretariat and any) s
5 654 M
(   assurances of licenses to be made available, or the result of an) s
5 643 M
(   attempt made to obtain a general license or permission for the use) s
5 632 M
(   of such proprietary rights by implementers or users of this) s
5 621 M
(   specification can be obtained from the IETF on-line IPR repository) s
5 610 M
(   at http://www.ietf.org/ipr.) s
5 599 M
(   The IETF invites any interested party to bring to its attention any) s
5 588 M
(   copyrights, patents or patent applications, or other proprietary) s
5 577 M
(   rights that may cover technology that may be required to implement) s
5 566 M
(   this standard.  Please address the information to the IETF at ietf-) s
5 555 M
(   ipr@ietf.org.) s
5 126 M
(Kuzmanovic                   Standards Track                   [Page 14]) s
_R
S
%%Trailer
%%Pages: 14
%%DocumentNeededResources: font Courier-Bold Courier 
%%EOF

PAFTECH AB 2003-20262026-04-22 08:54:10