From: Ivan (ivan@iip.net)
Date: Fri Jun 29 2007 - 08:07:14 ART
IMO mtu is unit wich L2-device can transmit through itself. L2 frame can be
1518 byte - ordinary ethernet, 1522 - dot1q frame, 1538 - isl encapsulation.
All this frames transmit only between directly connected interfaces. Switch
in any way process only 1500 payload (suppose it strip encapsulation header).
But if we need to transmit frame through switch system mtu must be changed.
For instance it can be dot1q-tunneling (+4 bytes), MPLS (+4*size_of_stack),
L2TP (+40 byte) .... (pppoe, gre, mpls te). To avoid fragmentation of frame
wich exceed 1500 byte there is need to increase MTU for L2 frames on switches
on all path of frame. After tune up switched path all frame will transmit
across ethernet segment intact.
Need to note that increase L2 MTU, don't inrease MTU for above layer. Ie using
gre on router also increase size of IP packet. If it will greater than 1500
byte it will fragmented by router although switch can transmit it through
switched path. To avoid fragmentation of packet need to increase IP-mtu on
each router on the path of tunnel.
The same for TAG MTU. Plain MPLS will have 1504 bytes length, MPLS VPN 1508
bytes. MPLS use tag-MTU parameter during fragmentation decision. To change
mtu for MPLS packet you can use mpls tag-mtu <value>.
On Friday 29 June 2007 14:24, Andy wrote:
> Maximum Ethernet size is 1518 bytes, is this 1518 bytes of payload, and the
> extras (headers, isl/dot1q) are "taken care" by the switch?
>
>
>
> So, when trunking with isl and we have a default MTU of 1500, are we
> sending 1530 bytes frame size?
>
> Same with dot1q and default MTU of 1500, are we sending 1504 bytes total?
>
>
>
> And why is system MTU configurable up to 1546 bytes when the max Ethernet
> (not gig) is only 1518?
>
>
>
> -A
>
> On 29/06/07, Brian Dennis <bdennis@internetworkexpert.com> wrote:
> > http://www.cisco.com/univercd/cc/td/doc/product/lan/cat3560/12225see/scg/
> >
> > System MTU
> >
> > The default system MTU for traffic on the switch is 1500 bytes. You can
> > configure Fast Ethernet ports to support frames larger than 1500 bytes by
> > using the system mtu global configuration command. You can configure
> > Gigabit
> > Ethernet ports to support frames larger than 1500 bytes by using the
> > system
> > mtu jumbo global configuration command. Because the IEEE 802.1Q tunneling
> > feature increases the frame size by 4 bytes when the metro tag is added,
> > you
> > must configure all switches in the service-provider network to be able to
> > process maximum frames by increasing the switch system MTU size to at
> > least
> > 1504 bytes. The maximum allowable system MTU for Gigabit Ethernet
> > interfaces
> > is 9000 bytes; the maximum system MTU for Fast Ethernet interfaces is
> > 1546 bytes.
> >
> > --
> >
> > Brian Dennis, CCIE4 #2210 (R&S/ISP-Dial/Security/SP)
> > bdennis@internetworkexpert.com
> >
> > Internetwork Expert, Inc.
> > http://www.InternetworkExpert.com
> > Toll Free: 877-224-8987
> > Direct: 775-745-6404 (Outside the US and Canada)
> >
> > On 6/28/07 10:51 PM, "Chris Proctor" <cproctor@sungardcollegis.com> wrote:
> > > Does anyone know where the 1504 byte is coming from? That doesn't even
> > > ring a bell as a standard MTU + encapsulation match (IP + MPLS shim
> > > maybe?) Is it just a code bug or what?
> > >
> > > -----Original Message-----
> > > From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
> > > John Gibson
> > > Sent: Sunday, June 24, 2007 2:46 PM
> > > To: Mark Mckillop (mmckillo); Antonio Soares; Radioactive Frog
> > > Cc: ccielab@groupstudy.com
> > > Subject: RE: RE: MTU Mismatch in OSPF
> > >
> > > I am thinking that if the switches have mtu 1504
> > > and routers have 1500, when the switches tries
> > > to initiate TCP connections (e.g. ftp download
> > > some image files) to the routers, the packet will
> > > , by default without extra protection mechanism,
> > > be dropped at some router.
> > >
> > > But TCP has a MSS that is around 1460 (with my
> > > linux box, and I hope this is the same in the catalyst
> > > switches) , which limits the IP packet size to
> > > 1500 when the OS tries to make packets. So FTP
> > > download will be fine.
> > >
> > > But I am thinking that TFTP download will fail.
> > > UDP does not have MSS.
> > >
> > > John
> > >
> > > --- "Mark Mckillop (mmckillo)" <mmckillo@cisco.com>
> > >
> > > wrote:
> > >> A nice option also is to do:
> > >>
> > >> system mtu routing 1500
> > >>
> > >> This will avoid you needing to use the mtu-ignore
> > >> command, although it
> > >> only works on the 3560's. It doesn't require a
> > >> reboot either. Just
> > >> ignoring something always feels like a bad thing
> > >> thing to do :)
> > >>
> > >> Mark.
> > >>
> > >> -----Original Message-----
> > >> From: nobody@groupstudy.com
> > >> [mailto:nobody@groupstudy.com] On Behalf Of
> > >> Antonio Soares
> > >> Sent: 23 June 2007 17:58
> > >> To: 'Radioactive Frog'; johngibson1541@yahoo.com
> > >> Cc: ccielab@groupstudy.com
> > >> Subject: RE: RE: MTU Mismatch in OSPF
> > >>
> > >> +++++++++++++++++++++++++++++++++++++++++++++++++++
> > >> The 3560 default is 1500:
> > >>
> > >> Switch#sh flash
> > >>
> > >> Directory of flash:/
> > >>
> > >> 2 -rwx 7963038 Mar 1 1993 00:57:27 +00:00
> > >> c3560-advipservicesk9-mz.122-25.SEE2.bin
> > >>
> > >> 32514048 bytes total (24549888 bytes free) Switch#
> > >> Switch# Switch#sh
> > >> flash
> > >> +++++++++++++++++++++++++++++++++++++++++++++++++++
> > >> Directory of flash:/
> > >>
> > >> 2 -rwx 7963038 Mar 1 1993 00:57:27 +00:00
> > >> c3560-advipservicesk9-mz.122-25.SEE2.bin
> > >>
> > >> 32514048 bytes total (24549888 bytes free)
> > >>
> > >> Switch#sh system mtu
> > >>
> > >> System MTU size is 1500 bytes
> > >> System Jumbo MTU size is 1500 bytes
> > >> Routing MTU size is 1500 bytes
> > >> Switch#
> > >> +++++++++++++++++++++++++++++++++++++++++++++++++++
> > >>
> > >> But the way, when you have an OSPF Mtu Mismatch,
> > >> where you do add the
> > >> "ip ospf mtu-ignore" command ?
> > >>
> > >> 1) The device with the lower Mtu.
> > >> 2) The device with the higher Mtu.
> > >> 3) Both devices.
> > >>
> > >>
> > >>
> > >>
> > >> -----Original Message-----
> > >> From: nobody@groupstudy.com
> > >> [mailto:nobody@groupstudy.com] On Behalf Of
> > >> Radioactive Frog
> > >> Sent: sabado, 23 de Junho de 2007 5:23
> > >> To: johngibson1541@yahoo.com
> > >> Cc: ccielab@groupstudy.com
> > >> Subject: Re: RE: MTU Mismatch in OSPF
> > >>
> > >> dats sound strange to me!
> > >> frog
> > >>
> > >>
> > >> On 6/23/07, johngibson1541@yahoo.com
> > >>
> > >> <johngibson1541@yahoo.com> wrote:
> > >>> I think 3550's MTU at 1504 is caused by a file
> > >>
> > >> flash:env_vars
> > >>
> > >>> If I remove this file, then reload it will become
> > >>
> > >> 1500.
> > >>
> > >>> My 3560 is a different story. It is 1504 by
> > >>
> > >> itself.
> > >>
> > >>> Strangely, univercd keeps talking about 3560 and
> > >>
> > >> 3550 default mtu
> > >>
> > >>> being 1500. They are just not saying the correct
> > >>
> > >> thing.
> > >>
> > >>> 3560 at 12.2(25) , 12.2(35) , and the latest
> > >>
> > >> 12.2(37) all make the
> > >>
> > >>> same mistake!
> > >
> > > ______________________________________________________________________
> > >
> > >>> _ Subscription information may be found at:
> > >>> http://www.groupstudy.com/list/CCIELab.html
> > >
> > > _______________________________________________________________________
> > >
> > >> Subscription information may be found at:
> > >> http://www.groupstudy.com/list/CCIELab.html
> > >
> > > _______________________________________________________________________
> > >
> > >> Subscription information may be found at:
> > >> http://www.groupstudy.com/list/CCIELab.html
> > >
> > > _______________________________________________________________________
> > >_ ____________
> > > Shape Yahoo! in your own image. Join our Network Research Panel today!
> > > http://surveylink.yahoo.com/gmrs/yahoo_panel_invite.asp?a=7
> > >
> > > _______________________________________________________________________
> > > Subscription information may be found at:
> > > http://www.groupstudy.com/list/CCIELab.html
> > >
> > > _______________________________________________________________________
> > > Subscription information may be found at:
> > > http://www.groupstudy.com/list/CCIELab.html
> >
> > _______________________________________________________________________
> > Subscription information may be found at:
> > http://www.groupstudy.com/list/CCIELab.html
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Sun Jul 01 2007 - 17:24:52 ART