Hi,
I do understand - but ultimately what I'm saying is that I'd prefer to
know that an IX switch will pass all frames asked of it, even if the
MTU nominally for IPv4 is 1500 bytes.
We set all ethernet switches internally to pass the maximum MTU
(usually 9180 bytes). However, our native MTU on routers is lower.
This way we never have to reconfigure or change it to support higher
MTU on the routers where it's a trivial interface change vs (for our
main vendor anyway) a reboot.
Although I can't think of an obvious application at the moment, it
maybe that at some point I do want to pass larger than 1500 bytes MTU
frames to a peer for some other reason than IPv4.
I'm not really that fussed though - I can't see a rapid adoption of
large frames at IXes for a while yet.
MMC
Raphael Ho wrote:
Matt,
Thanks for your comments.
The danger is when someone sets up for 1500 MTU and someone else sets
up for 9180 MTU. My switch would be happy to forward on the packets,
but the packet may be dropped in one direction.
It is imperative that everyone agrees with the same MTU (and agrees to
getting decent enough equipment to support this MTU) or it will cause
problems. That’s why I’m asking everybody on a public forum. ;-)
Regards,
Raf.
On 10/11/2008 3:35 PM, "Matthew Moyle-Croft" <mmc@internode.com.au>
wrote:
Hi Raphael,
My feeling is that 9180 byte MTUs would be nice, even if normally
unused. That way, if we and another peer bilaterally want to exchange
data with large MTUs we can.
However, 1500 byte MTUs, as you say, will be the norm for a while yet.
MMC
Raphael Ho wrote:
Jumbo Frames on IX Hi everybody,
Thanks for your feedback on my last email regarding v4/v6 on the
exchange. We’re rolling out the route-servers and IPv6 addresses on
Equinix Exchanges across Asia (Singapore, Tokyo, Sydney and HK), so if
anybody needs a port or an address, please feel free to let me know.
Now for another poll – on our new exchange in Hong Kong...
1. Is there value in enabling Jumbo Frames on the exchange?
2. If yes, what sized Jumbo frame do you want?
a) 1600 (baby giant)
b) 4470 (legacy MTU of FDDI, now default for POS links)
c) 9180 (RFC1209)
d) Other (please specify)
My personal concern is that there may be router/switch compatibilities
– some people’s equipment may not support jumbo frames of any kind and
cause potential issues (although chances are, most endpoint traffic is
still 1500bytes these days – maybe life will be different in a few
years when GPON brings fiber to everybody’s home). However, the world
will be so much more efficient with a larger MTU.
Just wanted to hear your feedback
Regards,
Raf.
_______________________________________________
sig-ix mailing list
sig-ix@lists.apnic.net
http://mailman.apnic.net/mailman/listinfo/sig-ix
|
