RE: RoE header and mapper thoughts: EtherType
That should be easy to achieve via subtype mechanism, similar to what was
done in 802.3, Clause 57
-----Original Message-----
From: AshwoodsmithPeter [mailto:Peter.AshwoodSmith@xxxxxxxxxx]
Sent: Monday, February 23, 2015 9:59 AM
To: Jouni Korhonen; Marek Hajduczenia
Cc: STDS-1904-3-TF@xxxxxxxxxxxxxxxxx
Subject: RE: RoE header and mapper thoughts: EtherType
I would agree. A single Ethertype should be sufficient for all ROE work.
We then need to carefully design the mechanisms for specyfing the
encapsulation that follows and what is control, what is data. Also there are
several different types of control when we get to the structure aware stuff.
Some control has to be delivered within a several hundred nanoseconds to
stay more or less synchronized with the data frames. Some can wait 10's of
micro seconds and some has a few milliseconds of tollerance.
Peter
________________________________________
From: stds-1904-3-tf@xxxxxxxx [stds-1904-3-tf@xxxxxxxx] on behalf of Jouni
Korhonen [jouni.korhonen@xxxxxxxxxxxx]
Sent: Monday, February 23, 2015 6:42 AM
To: Marek Hajduczenia
Cc: STDS-1904-3-TF@xxxxxxxxxxxxxxxxx
Subject: Re: RoE header and mapper thoughts: EtherType
Hi,
Ok Good.
- jouni
Sent from a smart phone.. Mind the typos..
> Marek Hajduczenia <marek.hajduczenia@xxxxxxxxx> kirjoitti 23.2.2015 kello
4.02:
>
> Jouni,
>
> I believe the agreement was that we do request one Ethertype to
> identify this new application type, but I do not believe there is
> anything that warrants a separate Ethertype for control traffic within RoE
tunnel.
>
> Marek
>
> -----Original Message-----
> From: Jouni Korhonen [mailto:jouni.korhonen@xxxxxxxxxxxx]
> Sent: Monday, February 23, 2015 12:50 AM
> To: Marek Hajduczenia; STDS-1904-3-TF@xxxxxxxxxxxxxxxxx
> Subject: RE: RoE header and mapper thoughts: EtherType
>
> Marek,
>
>> -----Original Message-----
>> From: Marek Hajduczenia [mailto:marek.hajduczenia@xxxxxxxxx]
>> Sent: Friday, February 20, 2015 4:41 PM
>> To: Jouni Korhonen; STDS-1904-3-TF@xxxxxxxxxxxxxxxxx
>> Subject: RE: RoE header and mapper thoughts: EtherType
>>
>> Jouni
>>
>> I think it is a bad idea. First, Ethertypes are a scarce resource and
>> will not be given away easily. Second, 802.3br should take care of
>> this
> problem.
>> I do not believe you need to signal special frame type via Ethertype
> [Jouni Korhonen]
>
> Do you no RoE EtherType at all or just not for RoE subtypes (like
> management)?
>
> - Jouni
>
>>
>> Marek
>>
>> -----Original Message-----
>> From: stds-1904-3-tf@xxxxxxxx [mailto:stds-1904-3-tf@xxxxxxxx] On
>> Behalf Of Jouni Korhonen
>> Sent: Friday, February 20, 2015 5:14 PM
>> To: STDS-1904-3-TF@xxxxxxxxxxxxxxxxx
>> Subject: RoE header and mapper thoughts: EtherType
>>
>> Folks,
>>
>> Some initial thoughts on reserving EtherType.. this is not topical
>> yet, though. My initial thinking was to have a single new EtherType
>> for everything RoE. However, recently I was thinking whether it would
>> make any sense to have two
>> EtherTypes: one for time critical packets (like AxC flows) and
>> another for less critical packets (like C&M flows etc).
>>
>> Opinions?
>>
>> - Jouni
>>
>> --
>> Jouni Korhonen, Ph.D, Associate Technical Director CTO Office,
>> Networking, Broadcom Corporation
>> O: +1-408-922-8135, M: +1-408-391-7160
>=