I think it is desirable for the fabric to be jumbo-frame compatible.

In general, jumbo frame support can be a good thing and we are fully jumbo-frame capable on our backbone.

I don't know of any downside to enabling jumbo frames on the fabric. Obviously enabling them on router interfaces requires the cooperation of all peers. While I'd like to see that happen, I'm guessing it's not as easy as one might hope.

Owen

On Jul 24, 2014, at 21:06 , Shaun Carlson <[log in to unmask]> wrote:

The discussion at the meeting is unfortunately missed in the briefness of the minutes.

The IX is configured to support jumbo frames and we felt we should keep it that way going forward.  This allows maximum flexibility to support possible applications of jumbo frames.

That being said, nobody is using them or has any concrete plans (that we could identify at the meeting) to use jumbo frames.  If a member has a specific application or desire to use jumbo frames, it's up to that member to work with whomever their peer is to work through the technicalities and concerns of that application.

s

On Thursday, July 24, 2014, Hannigan, Martin <[log in to unmask]> wrote:


Generally, the general IXP community has determined that the minimum exchange services configuration should be:

     http://www.open-ix.org/certification/ixp-technical-requirements/

If we all thought it was desired as a minimum configuration at an IXP, it would've made it in there IMHO. It didn't. If someone thinks all IXPs should do this, they should take a run at this and see if they can get the rest of the IXP community to support it (Not just Netnod and NASA).

We have the same issue as Andrew, and we have zero interest to go to an MTU of 9000.


Best,

-M<




On Jul 24, 2014, at 6:29 PM, "Koch, Andrew" <[log in to unmask])">[log in to unmask]> wrote:

> The UG 13 minutes make note that the exchange MTU is 9000 bytes.  Are other members configuring their routers for an MTU of 9000?
>
> I ask as TDS has not - we have stuck with the Ethernet standard 1500 bytes.  If anyone on the exchange has set an MTU of 9000 and tries to send jumbo traffic to us, it will not be accepted and no indication will be sent back to the originator. (10d in the linked document below)
>
> If no one (or few) has setup their routers for an MTU of 9000, but we as a community wish to enable this, I would recommend that we schedule a flag day so that we can minimize blackholing of traffic across the exchange.
>
> While only a draft, see http://tools.ietf.org/html/draft-mlevy-ixp-jumboframes-00 for further information on IXP Jumbo Frame operation and conversion.  For those that like presentations more than IETF drafts, Martin Levy (of Hurricane Electric at the time) presented this at RIPE 64 - https://ripe64.ripe.net/presentations/139-Hurricane_Electric_-_Martin_Levy_-_IX_Jumbo_Frames_-_RIPE64_EIX-WG_-_Slovenia.pdf
>
>
> Best Regards,
>
> Andy Koch
> TDS Telecom - IP Network Operations
> [log in to unmask])">[log in to unmask]
> Desk: 608-664-4694
> Cell: 608-616-0072
>
>
>
> ########################################################################
>
> To unsubscribe from the MICE-DISCUSS list, click the following link:
> ?SUBED1=MICE-DISCUSS&A=1



--

Shaun Carlson
Network Engineering Manager | 
Arvig
ph: (218) 346-8673 | em: [log in to unmask]" rel="noreferrer" target="_blank">[log in to unmask]



To unsubscribe from the MICE-DISCUSS list, click the following link:
http://lists.iphouse.net/cgi-bin/wa?SUBED1=MICE-DISCUSS&A=1




To unsubscribe from the MICE-DISCUSS list, click the following link:
http://lists.iphouse.net/cgi-bin/wa?SUBED1=MICE-DISCUSS&A=1