Linux-Bulgaria.ORG
навигация

 

начало

пощенски списък

архив на групата

семинари ...

документи

как да ...

 

 

Предишно писмо Следващо писмо Предишно по тема Следващо по тема По Дата По тема (thread)

Re: lug-bg:


  • Subject: Re: lug-bg:
  • From: vlk@xxxxxxxxxxxxxxxxx (Vesselin Kolev)
  • Date: Sun, 9 Feb 2003 17:21:55 +0200



-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Malko osobenosti sledvashti moia predishen posting...

  Vse pak dobre e mashinite ti, chrez koito pravish tunnela
da sa s moshten processor i poveche pamet. Razbira se ne
si misli za neshto poveche ot Celleron na 500 MHz i 512 MB RAM.
Tazi mosht, koiato spomenavam shte ti e nuzhna samo ako shte
pravish goliam traffic v tunnela i shte ima mnogo zaiavki.

Primerno, ako shte prekarvash 10 Mbps prez tunnela shte ti
stigne i edin Pentium MMX na 200 MHz s 256 MB RAM.. Viarno,
malko bavni shte sa dogovoarianiata (efecta mozhesh da go
vidish kato ustanovish sesia anagazhirashta goliama lenta ot
traffica i pingvash host v drugata mrezha - shte vidish kak ot
vremena 3-4 msec shte se kachish na 100-200, che i poveche).
Za da izbegnesh pone malko nepriatnite efekti si poigrai malko s ToS
markirovkata na paketite. Mozhe da ima poniakoga i zaguba
na paketi, tova obache e pri nalichieto na mnogo zaiavki za
preminavane v tunnela. Ako v mrezhata si imash liubiteli na
"Ping ot death" i dr takiva entusiasti sys selsko i poluseslko
vyzpitanie, napravi na dvata kraia na tunnela limit na goleminata
i na broia ICMP ping paketi za da smekchish efektite ot detskite
igri. No ne zabraniavai ICMP echo/request.. glupavo e.

Ako shte gradish mrezha s kapacitet ot 100 Mbps shte e dobre
da se podgotvish po-dobre otkym hardware (istinata e, che nikoga
niama da ia dokarash do proeknia kapacitet, no tova e dylga tema). 
Edin router ot
roda na Athlon na 750 MHz i 516 RAM kakyvto az izpolzvam,
shte opravi rabotata,
ako ne puskash na routerite X, Squid i java orientirani HTTPD-ta
(ne che i s tiah ne mozhe da se mine, no vseki iska idealna
mrezhova kartina). Opityt mi sochi, che s takava mashina 
mozhesh da vyrshish idealno rabota dori kato imash 3 izgradeni
tunela kym koito mashinata ti e svyrzana.

Napravo ti davam primer, zashtoto v momenta minavam prez
tunnel. V momenta traffica prez tunnela e 17.29 Mbps. Tunnela
e izgraden m/u dve Intelski 10/100 Mbps. Tunnela preminava 
prez 4 mashini i 2 switcha. Vryzkata ot edinia krai na tunnela
do drugia e vyv FDX Mode.

Eto ti rezultata ot edin traceroute do free.techno-link.com

[vlk@newton vlk]$ traceroute free.techno-link.com
traceroute to free.techno-link.com (212.91.161.253), 30 hops max, 38 byte 
packets
 1  nat-router-to-digsys.backbone-2.vpn.lcpe.uni-sofia.bg (192.168.100.1)  
0.353 ms  0.201 ms  0.167 ms
 2  eth-out.backbone-1.lcpe.uni-sofia.bg (62.44.103.2)  2.477 ms  2.303 ms  
3.261 ms
 3  Sofia4.BG.EU.net (193.68.0.198)  3.089 ms  3.654 ms  2.801 ms
 4  Sofia5.BG.EU.net (193.68.0.172)  4.027 ms  5.202 ms  3.442 ms
 5  62.176.102.110 (62.176.102.110)  4.773 ms  14.387 ms  4.604 ms
 6  free.techno-link.com (212.91.161.253)  5.626 ms  6.668 ms  6.258 ms
[vlk@newton vlk]$ 

Tunnela zapochva pri pyrvia host i svyrshva pri vtoria. 

Eto ti ping rezultati bez i s tunnel:

- --> s tunnel

[vlk@newton vlk]$ ping free.techno-link.com
PING free.techno-link.com (212.91.161.253) from 192.168.100.111 : 56(84) bytes 
of data.
64 bytes from free.techno-link.com (212.91.161.253): icmp_seq=0 ttl=58 
time=8.061 msec
64 bytes from free.techno-link.com (212.91.161.253): icmp_seq=1 ttl=58 
time=7.301 msec
64 bytes from free.techno-link.com (212.91.161.253): icmp_seq=2 ttl=58 
time=5.280 msec
64 bytes from free.techno-link.com (212.91.161.253): icmp_seq=3 ttl=58 
time=5.559 msec

- --> bez tunnel

[vlk@lcpe vlk]$ ping free.techno-link.com
PING free.techno-link.com (212.91.161.253) from 192.168.100.111 : 56(84) bytes 
of data.
64 bytes from free.techno-link.com (212.91.161.253): icmp_seq=0 ttl=58 
time=4.061 msec
64 bytes from free.techno-link.com (212.91.161.253): icmp_seq=1 ttl=58 
time=5.301 msec
64 bytes from free.techno-link.com (212.91.161.253): icmp_seq=2 ttl=58 
time=4.280 msec
64 bytes from free.techno-link.com (212.91.161.253): icmp_seq=3 ttl=58 
time=3.559 msec

Samo kato iliustracia na vyrzmozhnostite, koito mozhe da ti predlozhi edin 
tunnel... eto ti i edin primer s tunnel izgraden po mezhdunarodna linia:

[vlk@newton vlk]$ traceroute www.ripe.net
traceroute to peach.ripe.net (193.0.0.203), 30 hops max, 38 byte packets
 1  nat-router-to-digsys.backbone-3.vpn.lcpe.uni-sofia.bg (192.168.100.40)  
0.763 ms  0.449 ms  0.170 ms
 2  Mandrake.nat-lan.lcpe.pip.digsys.bg (193.68.191.198)  0.425 ms  0.318 ms  
0.256 ms
 3  XXX.XXX.XXX.XXX 416.033 ms  427.955 ms  417.664 ms
 4  Amsterdam1.ripe.net (193.148.15.68)  430.920 ms  414.864 ms  417.717 ms
 5  peach.ripe.net (193.0.0.203)  429.204 ms  437.216 ms  420.556 ms
[vlk@newton vlk]$ 

kato tunnela otiva do edin router na DTAG (narochno sym go skril i ne sym
opisal imeto mu)v Holland, koito napravo ima izlaz
kym mrezhata na RIPE, no toi e pusnat s IPSec, .t.e. ne e syvsem po nashia
prmer de... no pak stava. Tunnela zapochva pri vtoria host i zavyrshva pri
3-tia.

Eto ti razlikata vyv vremenata za otgovor na ping s i bez tunnel:

- --> bez tunnel:
 [vlk@lcpe vlk]$ ping www.ripe.net
PING peach.ripe.net (193.0.0.203) from 192.168.100.111 : 56(84) bytes of data.
64 bytes from peach.ripe.net (193.0.0.203): icmp_seq=0 ttl=244 time=324.385 
msec
64 bytes from peach.ripe.net (193.0.0.203): icmp_seq=1 ttl=244 time=311.658 
msec
64 bytes from peach.ripe.net (193.0.0.203): icmp_seq=2 ttl=244 time=322.182 
msec
64 bytes from peach.ripe.net (193.0.0.203): icmp_seq=3 ttl=244 time=322.793 
msec

- --> s tunnel:
[vlk@newton vlk]$ ping www.ripe.net
PING peach.ripe.net (193.0.0.203) from 192.168.100.111 : 56(84) bytes of data.
64 bytes from peach.ripe.net (193.0.0.203): icmp_seq=0 ttl=244 time=426.256 
msec
64 bytes from peach.ripe.net (193.0.0.203): icmp_seq=1 ttl=244 time=423.667 
msec
64 bytes from peach.ripe.net (193.0.0.203): icmp_seq=2 ttl=244 time=417.576 
msec
64 bytes from peach.ripe.net (193.0.0.203): icmp_seq=3 ttl=244 time=421.334 
msec

No tozi kanal e 2Mbps, a ne 100 Mbps i tam zavisia ot mnogo nekontrolirani
ot men factori, no obshto vzeto mozhe da se dobie predstava za efectite.

Nakraia shte ti pokazha kakvo stava kato se pusne ICMP ping s golemi paketi
v tunnel-a izgraden po 100 Mbps linia:

[vlk@velociraptor vlk]$ ping -s 34568 193.68.191.193
PING 193.68.191.193 (193.68.191.193) from 192.168.100.111 : 34560(34588) bytes 
of data.
34568 bytes from 192.168.100.11: icmp_seq=0 ttl=128 time=128.402 msec
34568 bytes from 192.168.100.11: icmp_seq=1 ttl=128 time=214.363 msec
34568 bytes from 192.168.100.11: icmp_seq=2 ttl=128 time=169.379 msec
34568 bytes from 192.168.100.11: icmp_seq=3 ttl=128 time=150.332 msec

za sravnenie normalnite paketi minavat taka::

[vlk@velociraptor vlk]$ ping 193.68.191.193
PING 193.68.191.193 (193.68.191.193) from 192.168.100.111 : 34560(34588) bytes 
of data.
64 bytes from 192.168.100.11: icmp_seq=0 ttl=128 time=861 usec
64 bytes from 192.168.100.11: icmp_seq=1 ttl=128 time=1.054 msec
64 bytes from 192.168.100.11: icmp_seq=2 ttl=128 time=1.157 msec
64 bytes from 192.168.100.11: icmp_seq=3 ttl=128 time=255 usec

 Tova e.. mislia, che se kazaha mnogo neshto po vyrposa i se pokazaha
primeri

  Tova e ot men

   Pozdravi
      Vesselin Kolev
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)

iD8DBQE+RnIZ+48lZPXaa+MRAgeaAKCTKRAMzBeL8X33AbFTBF6pbcm7mACdF8e4
jrtUfLGmBxR3yLgPAI5zD20=
=lrHG
-----END PGP SIGNATURE-----

============================================================================
A mail-list of Linux Users Group - Bulgaria (bulgarian linuxers).
http://www.linux-bulgaria.org - Hosted by Internet Group Ltd. - Stara Zagora
To unsubscribe: http://www.linux-bulgaria.org/public/mail_list.html
============================================================================


  • Относно:
    • lug-bg:
      • Изпратено от: todor.lazarov@xxxxxxxxxxxxxxxx (Todor Lazarov)

 

наши приятели

 

линукс за българи
http://linux-bg.org

FSA-BG
http://fsa-bg.org

OpenFest
http://openfest.org

FreeBSD BG
http://bg-freebsd.org

KDE-BG
http://kde.fsa-bg.org/

Gnome-BG
http://gnome.cult.bg/

проект OpenFMI
http://openfmi.net

NetField Forum
http://netField.ludost.net/forum/

 

 

Linux-Bulgaria.ORG

Mailing list messages are © Copyright their authors.