The Ethernet standards (Blue Book and IEEE 802.3) say that the maximum
Ethernet frame size is 1518 bytes, counting the dest and src addresses,
type/length field, and FCS. But, theoretically, there's no reason it
couldn't be bigger, and some interfaces accept bigger frames. They have to
for ISL, as you mentioned.
I asked Bob Metcalfe where 1518 bytes comes from. Seriously, I met him at a
party many years ago and hit him with this question. His answer, "Hmm, well
I really don't know!"
Some sort of maximum is necessary, of course, in order to assure fairness.
The myths about the maximum having something to do with making CSMA/CD work
correctly are inaccurate, (not that anyone said that in this thread). The
myths are a mistake by analogy. The minimum size frame is necessary for
CSMA/CD to work correctly.
At 09:22 AM 12/4/00, Tony van Ree wrote:
>I was under the impression that an ethernet frame had a maximum size of
>1500 bytes. This did not include the addresses or the FCS. To increase
>the MTU beyond that size would indeed create a giant on ethernet. As it
>is you will find some devices give an error message when including an ISL
>header. (A gig port on a 2984 is a good example) In the case of the 2984
>the data still goes through ok but when you get a packet over 1484 bytes
>an whack it through the port the error rate rises. Anything small is ok.
>It does seem to me that the basic ethernet standards define the max MTU
>I could however have mis understood all these years.
>On Saturday, December 02, 2000 at 12:25:13 AM, Kevin Wigle wrote:
> > there is a "mtu" command.
> > In a somewhat similar situation - I had built a circuit using lan emulation
> > that terminated on a bvi on a 7505.
> > The bvi had an ip address and placed into an OSPF area.
> > I'm not aware of the defaults with a router that has both ATM and ethernet
> > but I didn't take notice of the mtu that the bvi had - I assumed that
> it was
> > just like an ethernet interface........
> > On the distant end of the lan emulation was an ordinary router with an
> > ethernet interface and it was also configured into the OSPF area.
> > However, OSPF didn't work. debugging showed that an adjacency wasn't being
> > formed.
> > Turning up yet more debugging - finally an error about "mtu size not equal"
> > was noticed.
> > sure enough, the bvi had the ATM mtu of 4470..........
> > using the mtu command on the bvi interface "mtu 1500", the adjacency formed
> > immediately and all was well.
> > so, my guess would be that a fast ethernet probably has the same "mtu"
> > command, try it out.
> > Kevin Wigle
> > ----- Original Message -----
> > From: "Darren Ward" <[EMAIL PROTECTED]>
> > To: <[EMAIL PROTECTED]>
> > Sent: Friday, 01 December, 2000 22:02
> > Subject: Fast Ethernet MTU Size
> > > Hi All,
> > >
> > > Is it at all possible to raise the MTU size on Fast Ethernet?
> > >
> > > I'll give you all a scenario....
> > >
> > > ATM on one side with an MTU of 4470, fast ethernet connecting the two
> > > routers, Gigabit Ethernet the other side with an MTU of 4470.
> > >
> > > How can I raise the MTU of a Full Duplex Fast Ethernet Connection above
> > > 1500?
> > >
> > > I assume it's not possible when going through switched as they will show
> > > every packet as a giant but in a router to router cross-over connection
> > > I was hoping there was some way to keep the MTU static across the path
> > > rather than force the routers to fragment and re-assemble (of course the
> > > destination re-assembles).
> > >
> > > Darren Ward
> > > CCNP, CCDP, CCIE Wannabee
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]