1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646
|
#
# many_tcp.tcl
# $Id: many_tcp.tcl,v 1.19 2010/04/03 20:40:14 tom_henderson Exp $
#
# Copyright (c) 1998 University of Southern California.
# All rights reserved.
#
# Redistribution and use in source and binary forms are permitted
# provided that the above copyright notice and this paragraph are
# duplicated in all such forms and that any documentation, advertising
# materials, and other materials related to such distribution and use
# acknowledge that the software was developed by the University of
# Southern California, Information Sciences Institute. The name of the
# University may not be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED "AS IS" AND WITHOUT ANY EXPRESS OR IMPLIED
# WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF
# MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
#
proc usage {} {
puts stderr {usage: ns rbp_simulation.tcl [options]
This simulation demonstrates large numbers of TCP flows.
Comments/bugs to John Heidemann <johnh@isi.edu>.
Currently this simulation appears to be memory limited for most
practical purposes. Each active flow consumes ~100KB of memory (very
roughly).
Why is a flow so expensive? Primarly because We allocate a separate
pair of nodes and links per flow. (I don't know how to do otherwise
and preserve some reasonable statement about the independence of
randomly selected client RTTs).
How can performance be improved?
- use multiple concurrent flows per node (should give much better
memory performance at the cost of node RTT independence)
- use session-level simulation techniques (see ``Enabling Large-scale
simulations: selective abstraction approach to the study of multicast
protocols'' by Huang, Estrin, and Heidemann in MASCOTS '98).
- improve the efficiency of the underlying representations (there are
some ways to optimize some otcl common cases) (this work is in progress)
- use higher level representations of TCP (in progress)
Options (specify with -OPTIONNAME OPTIONVALUE):
}
global raw_opt_info
puts stderr $raw_opt_info
exit 1
}
global raw_opt_info
set raw_opt_info {
# how long to run the sim?
duration 30
# initilization: just start n clients at time 0
# NEEDSWORK: a more realistic ramp-up model
# or some kind of autodetection on when we've reached
# steady state would be nice.
initial-client-count 10
#
# BASIC TOPOLOGY:
#
# (The basic n clients on the left and right going through a
# bottleneck.)
#
# cl_1 cr_1
# ... ---- bottleneck_left ---- bottleneck_right --- ...
# cl_n cr_n
#
# node-number 0 specifies a new pair of nodes
# on the left and right for each new client
node-number 0
# NEEDSWORK:
# The number of agents attached to a node cannot exceed
# the port-field length (255 bits). There is currently
# no check or warning message for this.
#
# Currently all data traffic flows left-to-right.
# NEEDSWORK: relax this assumption (but Poduri and Nichols
# I-D suggests that relaxing it won't change things much).
#
#
# CLIENT TRAFFIC MODEL:
#
# arrival rate per second (arrival is poisson)
client-arrival-rate 1
# Currently clients are either mice or elephants.
# NEEDSWORK: should better model http-like traffic patterns.
# In particular, netscape's 4-connection model makes
# a *big* difference in traffic patterns
# and is not currently modeled at all.
client-mouse-chance 90
client-mouse-packets 10
client-elephant-packets 100
# For traffic in the reverse direction.
client-reverse-chance 0
# Pkt size in bytes.
# NEEDSWORK: should check that everything is uniformly
# specified (router queues are in packets of 1000B length?).
client-pkt-size 576
#
# CLIENT NETWORK CONNECTION:
#
client-bw 56kb
# client-server rtt is uniform over this range (currently)
# NEEDSWORK: does this need to be modeled more accurately?
client-delay random
client-delay-range 100ms
client-queue-method DropTail
# Insure that client routers are never a bottleneck.
client-queue-length 100
#
# CLIENT/SERVER TCP IMPLEMENTATION:
#
# NEEDSWORK: should add HTTP model over TCP.
source-tcp-method TCP/Reno
sink-ack-method TCPSink/DelAck
# Set init-win to 1 for initial windows of size 1.
# Set init-win to 10 for initial windows of 10 packets.
# Set init-win to 0 for initial windows per internet-draft.
init-win 1
#
# BOTTLENECK LINK MODEL:
#
bottle-bw 10Mb
bottle-delay 4ms
bottle-queue-method RED
# bottle-queue-length is either in packets or
# is "bw-delay-product" which does the currently
# expected thing.
bottle-queue-length bw-delay-product
#
# OUTPUT OPTIONS:
#
graph-results 0
# Set graph-scale to 2 for "rows" for each flow.
graph-scale 1
graph-join-queueing 1
gen-map 0
mem-trace 0
print-drop-rate 0
debug 1
title none
# set test-suite to write the graph to opts(test-suite-file)
test-suite 0
test-suite-file temp.rands
# Random number seed; default is 0, so ns will give a
# diff. one on each invocation.
ns-random-seed 0
# Animation options; complete traces are useful
# for nam only, so do those only when a tracefile
# is being used for nam
# Set trace-filename to "none" for no tracefile.
trace-filename out
trace-all 0
namtrace-some 0
namtrace-all 0
# Switch to generate the nam tcl file from here
# itself
nam-generate-cmdfile 0
}
Class Main
Main instproc default_options {} {
global opts opt_wants_arg raw_opt_info
set cooked_opt_info $raw_opt_info
while {$cooked_opt_info != ""} {
if {![regexp "^\[^\n\]*\n" $cooked_opt_info line]} {
break
}
regsub "^\[^\n\]*\n" $cooked_opt_info {} cooked_opt_info
set line [string trim $line]
if {[regexp "^\[ \t\]*#" $line]} {
continue
}
if {$line == ""} {
continue
} elseif [regexp {^([^ ]+)[ ]+([^ ]+)$} $line dummy key value] {
set opts($key) $value
set opt_wants_arg($key) 1
} else {
set opt_wants_arg($key) 0
# die "unknown stuff in raw_opt_info\n"
}
}
}
Main instproc process_args {av} {
global opts opt_wants_arg
$self default_options
for {set i 0} {$i < [llength $av]} {incr i} {
set key [lindex $av $i]
if {$key == "-?" || $key == "--help" || $key == "-help" || $key == "-h"} {
usage
}
regsub {^-} $key {} key
if {![info exists opt_wants_arg($key)]} {
puts stderr "unknown option $key";
usage
}
if {$opt_wants_arg($key)} {
incr i
set opts($key) [lindex $av $i]
} else {
set opts($key) [expr !opts($key)]
}
}
}
proc my-duplex-link {ns n1 n2 bw delay queue_method queue_length} {
global opts
$ns duplex-link $n1 $n2 $bw $delay $queue_method
$ns queue-limit $n1 $n2 $queue_length
$ns queue-limit $n2 $n1 $queue_length
}
Main instproc init_network {} {
global opts fmon
# nodes
# build right to left
$self instvar bottle_l_ bottle_r_ cs_l_ cs_r_ ns_ cs_count_ ns_ clients_started_ clients_finished_
#
# Figure supported load.
#
set expected_load_per_client_in_bps [expr ($opts(client-mouse-chance)/100.0)*$opts(client-mouse-packets)*$opts(client-pkt-size)*8 + (1.0-$opts(client-mouse-chance)/100.0)*$opts(client-elephant-packets)*$opts(client-pkt-size)*8]
if {$opts(debug)} {
set max_clients_per_second [expr [$ns_ bw_parse $opts(bottle-bw)]/$expected_load_per_client_in_bps]
puts [format "maximum clients per second: %.3f" $max_clients_per_second]
}
# Compute optimal (?) bottleneck queue size
# as the bw-delay product.
if {$opts(bottle-queue-length) == "bw-delay-product"} {
set opts(bottle-queue-length) [expr ([$ns_ bw_parse $opts(bottle-bw)] * ([$ns_ delay_parse $opts(bottle-delay)] + [$ns_ delay_parse $opts(client-delay-range)]) + $opts(client-pkt-size)*8 - 1)/ ($opts(client-pkt-size) * 8)]
puts "optimal bw queue size: $opts(bottle-queue-length)"
}
# Do our own routing with expanded addresses (21 bits nodes).
# (Basic routing limits us to 128 nodes == 64 clients).
$ns_ rtproto Manual
$ns_ set-address-format expanded
# set up the bottleneck
set bottle_l_ [$ns_ node]
set bottle_r_ [$ns_ node]
my-duplex-link $ns_ $bottle_l_ $bottle_r_ $opts(bottle-bw) $opts(bottle-delay) $opts(bottle-queue-method) $opts(bottle-queue-length)
if {$opts(print-drop-rate)} {
set slink [$ns_ link $bottle_l_ $bottle_r_]
set fmon [$ns_ makeflowmon Fid]
$ns_ attach-fmon $slink $fmon
}
# Bottlenecks need large routing tables.
# [$bottle_l_ set classifier_] resize 511
# [$bottle_r_ set classifier_] resize 511
# Default routes to the other.
[$bottle_l_ get-module "Manual"] add-route-to-adj-node -default $bottle_r_
[$bottle_r_ get-module "Manual"] add-route-to-adj-node -default $bottle_l_
# Clients are built dynamically.
set cs_count_ 0
set clients_started_ 0
set clients_finished_ 0
}
# create a new pair of end nodes
Main instproc create_client_nodes {node} {
global opts
$self instvar bottle_l_ bottle_r_ cs_l_ cs_r_ sources_ cs_count_ ns_ rng_
set now [$ns_ now]
set cs_l_($node) [$ns_ node]
set cs_r_($node) [$ns_ node]
# Set delay.
set delay $opts(client-delay)
if {$delay == "random"} {
set delay [$rng_ exponential [$ns_ delay_parse $opts(client-delay-range)]]
}
# Now divide the delay into the two haves and set up the network.
set ldelay [$rng_ uniform 0 $delay]
set rdelay [expr $delay - $ldelay]
my-duplex-link $ns_ $cs_l_($node) $bottle_l_ $opts(client-bw) $ldelay $opts(client-queue-method) $opts(client-queue-length)
my-duplex-link $ns_ $cs_r_($node) $bottle_r_ $opts(client-bw) $rdelay $opts(client-queue-method) $opts(client-queue-length)
# Add routing in all directions
[$cs_l_($node) get-module "Manual"] add-route-to-adj-node -default $bottle_l_
[$cs_r_($node) get-module "Manual"] add-route-to-adj-node -default $bottle_r_
[$bottle_l_ get-module "Manual"] add-route-to-adj-node $cs_l_($node)
[$bottle_r_ get-module "Manual"] add-route-to-adj-node $cs_r_($node)
if {$opts(debug)} {
# puts "t=[format %.3f $now]: node pair $node created"
# puts "delay $delay ldelay $ldelay"
}
}
# Get the number of the node pair
Main instproc get_node_number { client_number } {
global opts
if {$opts(node-number) > 0} {
set node [expr $client_number % $opts(node-number)]
} else {
set node $client_number
}
return $node
}
# return the client index
Main instproc create_a_client {} {
global opts
$self instvar cs_l_ cs_r_ sources_ cs_count_ ns_ rng_
# Get the client number for the new client.
set now [$ns_ now]
set i $cs_count_
incr cs_count_
set node $i
if {[expr $i % 100] == 0} {
puts "t=[format %.3f $now]: client $i created"
}
# Get the source and sink nodes.
if {$opts(node-number) > 0} {
if {$node < $opts(node-number) } {
$self create_client_nodes $node
} else {
set node [$self get_node_number $i]
}
} else {
$self create_client_nodes $node
}
if {$opts(debug)} {
# puts "t=[format %.3f $now]: client $i uses node pair $node"
}
# create sources and sinks in both directions
# (actually, only one source per connection, for now)
if {[$rng_ integer 100] < $opts(client-reverse-chance)} {
set sources_($i) [$ns_ create-connection-list $opts(source-tcp-method) $cs_r_($node) $opts(sink-ack-method) $cs_l_($node) $i]
} else {
set sources_($i) [$ns_ create-connection-list $opts(source-tcp-method) $cs_l_($node) $opts(sink-ack-method) $cs_r_($node) $i]
}
[lindex $sources_($i) 0] set maxpkts_ 0
[lindex $sources_($i) 0] set packetSize_ $opts(client-pkt-size)
# Set up a callback when this client ends.
[lindex $sources_($i) 0] proc done {} "$self finish_a_client $i"
if {$opts(debug)} {
# puts "t=[$ns_ now]: client $i created"
}
return $i
}
#
# Make a batch of clients to amortize the cost of routing recomputation
# (actually no longer improtant).
#
Main instproc create_some_clients {} {
global opts
$self instvar idle_clients_ ns_ cs_count_
set now [$ns_ now]
set step 16
if {$opts(debug)} {
puts "t=[format %.3f $now]: creating clients $cs_count_ to [expr $cs_count_ + $step - 1]"
}
for {set i 0} {$i < $step} {incr i} {
lappend idle_clients_ [$self create_a_client]
}
# debugging:
# puts "after client_create:"
# $ns_ gen-map
# $self instvar bottle_l_ bottle_r_
# puts "bottle_l_ classifier_:"
# [$bottle_l_ set classifier_] dump
# puts "bottle_r_ classifier_:"
# [$bottle_r_ set classifier_] dump
}
Main instproc start_a_client {} {
global opts
$self instvar idle_clients_ ns_ sources_ rng_ source_start_ source_size_ clients_started_
set i ""
set now [$ns_ now]
# can we reuse a dead client?
if {![info exists idle_clients_]} {
set idle_clients_ ""
}
while {$idle_clients_ == ""} {
$self create_some_clients
}
set i [lindex $idle_clients_ 0]
set idle_clients_ [lrange $idle_clients_ 1 end]
# Reset the connection.
[lindex $sources_($i) 0] reset
[lindex $sources_($i) 1] reset
# Start traffic for that client.
if {[$rng_ integer 100] < $opts(client-mouse-chance)} {
set len $opts(client-mouse-packets)
} else {
set len $opts(client-elephant-packets)
}
[lindex $sources_($i) 0] advanceby $len
set source_start_($i) $now
set source_size_($i) $len
if {$opts(debug)} {
# puts "t=[$ns_ now]: client $i started, ldelay=[format %.6f $ldelay], rdelay=[format %.6f $rdelay]"
puts "t=[format %.3f $now]: client $i started"
}
incr clients_started_
}
Main instproc finish_a_client {i} {
global opts
$self instvar ns_ idle_clients_ source_start_ source_size_ clients_finished_
set now [$ns_ now]
if {$opts(debug)} {
set delta [expr $now - $source_start_($i)]
puts "t=[format %.3f $now]: client $i finished ($source_size_($i) pkts, $delta s)"
}
lappend idle_clients_ $i
incr clients_finished_
}
Main instproc schedule_continuing_traffic {} {
global opts
$self instvar ns_ rng_
$self start_a_client
# schedule the next one
set next [expr [$ns_ now]+([$rng_ exponential]/$opts(client-arrival-rate))]
if {$opts(debug)} {
# puts "t=[$ns_ now]: next continuing traffic at $next"
}
$ns_ at $next "$self schedule_continuing_traffic"
}
Main instproc schedule_initial_traffic {} {
global opts
# Start with no pending clients.
$self instvar idle_clients_
# Start initial clients.
for {set i 0} {$i < $opts(initial-client-count)} {incr i} {
$self start_a_client
}
}
Main instproc open_trace { stop_time } {
global opts
$self instvar ns_ trace_file_ nam_trace_file_ trace_filename_
set trace_filename_ $opts(trace-filename)
exec rm -f "$trace_filename_.tr"
set trace_file_ [open "$trace_filename_.tr" w]
set stop_actions "close $trace_file_"
if {$opts(namtrace-some) || $opts(namtrace-all)} {
exec rm -f "$trace_filename_.nam"
set nam_trace_file_ [open "$trace_filename_.nam" w]
set $stop_actions "$stop_actions; close $nam_trace_file_"
} else {
set nam_trace_file_ ""
}
$ns_ at $stop_time "$stop_actions; $self finish"
return "$trace_file_ $nam_trace_file_"
}
# There seems to be a problem with the foll function, so quit plotting
# with -a -q, use just -a.
Main instproc finish {} {
global opts fmon PERL
$self instvar trace_filename_ ns_ cs_count_ clients_started_ clients_finished_
puts "total clients started: $clients_started_"
puts "total clients finished: $clients_finished_"
if {$opts(print-drop-rate)} {
set drops [$fmon set pdrops_]
set pkts [$fmon set parrivals_]
puts "total_drops $drops total_packets $pkts"
set droprate [expr 100.0*$drops / $pkts ]
puts [format "drop_percentage %7.4f" $droprate]
}
if {$opts(trace-filename) != "none"} {
set title $opts(title)
set flow_factor 1
if {$opts(graph-scale) == "2"} {
set flow_factor 100
}
# Make sure that we run in place even without raw2xg in our path
# (for the test suites).
set raw2xg raw2xg
if [file exists ../../bin/raw2xg] {
set raw2xg ../../bin/raw2xg
}
set raw2xg_opts ""
if {$opts(graph-join-queueing)} {
set raw2xg_opts "$raw2xg_opts -q"
}
# always run raw2xg because maybe we need the output
set cmd "$raw2xg -a $raw2xg_opts -n $flow_factor < $trace_filename_.tr >$trace_filename_.xg"
eval "exec $PERL $cmd"
if {$opts(graph-results)} {
if {$opts(graph-join-queueing)} {
exec xgraph -t $title < $trace_filename_.xg &
} else {
exec xgraph -tk -nl -m -bb -t $title < $trace_filename_.xg &
}
}
if {$opts(test-suite)} {
exec cp $trace_filename_.xg $opts(test-suite-file)
}
# exec raw2xg -a < out.tr | xgraph -t "$opts(server-tcp-method)" &
}
if {$opts(mem-trace)} {
$ns_ clearMemTrace
}
exit 0
}
Main instproc trace_stuff {} {
global opts
$self instvar bottle_l_ bottle_r_ ns_ trace_file_ nam_trace_file_
$self open_trace $opts(duration)
if {$opts(trace-all)} {
$ns_ trace-all $trace_file_
}
if {$opts(namtrace-all)} {
$ns_ namtrace-all $nam_trace_file_
} elseif {$opts(namtrace-some)} {
# xxx
$bottle_l_ dump-namconfig
$bottle_r_ dump-namconfig
[$ns_ link $bottle_l_ $bottle_r_] dump-namconfig
$ns_ namtrace-queue $bottle_l_ $bottle_r_ $nam_trace_file_
$ns_ namtrace-queue $bottle_r_ $bottle_l_ $nam_trace_file_
}
# regular tracing.
# trace left-to-right only
$ns_ trace-queue $bottle_l_ $bottle_r_ $trace_file_
$ns_ trace-queue $bottle_r_ $bottle_l_ $trace_file_
# Currently tracing is somewhat broken because
# of how the plumbing happens.
}
Main instproc init {av} {
global opts
$self process_args $av
$self instvar ns_
set ns_ [new Simulator]
# Seed random no. generator; ns-random with arg of 0 heuristically
# chooses a random number that changes on each invocation.
$self instvar rng_
set rng_ [new RNG]
$rng_ seed $opts(ns-random-seed)
$rng_ next-random
$self init_network
if {$opts(trace-filename) == "none"} {
$ns_ at $opts(duration) "$self finish"
} else {
$self trace_stuff
}
# xxx: hack (next line)
# $self create_some_clients
$ns_ at 0 "$self schedule_initial_traffic"
if {$opts(client-arrival-rate) != 0} {
$ns_ at 0 "$self schedule_continuing_traffic"
}
if {$opts(gen-map)} {
$ns_ gen-map
}
Agent/TCP set syn_ true
Agent/TCPSink set SYN_immediate_ack_ false ; # Default changed to true 2010/02
Agent/TCP set delay_growth_ true
Agent/TCP set windowInit_ 1
Agent/TCP set windowInitOption_ 1
if {$opts(init-win) == "0"} {
Agent/TCP set windowInitOption_ 2
} elseif {$opts(init-win) == "10"} {
Agent/TCP set windowInitOption_ 1
Agent/TCP set windowInit_ 10
} elseif {$opts(init-win) == "20"} {
Agent/TCP set windowInitOption_ 1
Agent/TCP set windowInit_ 20
puts "init-win 20"
}
$ns_ run
}
global in_test_suite
if {![info exists in_test_suite]} {
global $argv
new Main $argv
}
|