[c-nsp] general questions regarding MTU

Gert Doering gert at greenie.muc.de
Tue Apr 5 16:19:44 EDT 2011


Hi,

On Tue, Apr 05, 2011 at 03:57:21PM +0300, Martin T wrote:
> 1) as I understand, frames larger than MTU are automatically
> fragmented by switch. I understand, that it takes some processing
> power and probably time as well(latency increases), but are there any
> other disadvantages if fragmentation occurs?

No.  Switches do not fragment.  They will just drop a bigger-than-1500
packet coming in on some port if the egress port cannot handle that.

*Routers* will fragment (and it will cost).

> 2) when we talk about MTU(standard Ethernet v2 frame 1500B, jumbo
> frames 1500B - 9000B etc), then we don't count in L2 header? So for
> example 802.1qinq frame including multiple VLAN tags, srcMAC, dstMAC,
> EtherType and CRC with payload less 1500B is still counted as regular
> Ethernet v2 MTU?

Yes.  (If you transport these over MPLS, for example, you do have to
take them into account with 1518 or 1522 bytes)

> 3) on the image, all the switches have "system mtu 1552". Am I
> correct, that "system mtu" setting only affects 10BASE-T and
> 100BASE-TX ports? And this 1552B does not include the L2 header,
> right?

No idea how the MXes or 4500s do that.  

On a 6500, you can (and have to) enable jumbo MTU for each individual port.

But yes, you'll need to enable jumbo mtu along the whole path.

gert

-- 
USENET is *not* the non-clickable part of WWW!
                                                           //www.muc.de/~gert/
Gert Doering - Munich, Germany                             gert at greenie.muc.de
fax: +49-89-35655025                        gert at net.informatik.tu-muenchen.de
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 305 bytes
Desc: not available
URL: <https://puck.nether.net/pipermail/cisco-nsp/attachments/20110405/d711f4d1/attachment.pgp>


More information about the cisco-nsp mailing list