On Friday, January 28, 2011 11:25 AM, Ron Cohen [[email protected]] wrote:
> I don't think it is a good idea. In my opinion we should not / can not 
> specify that packets with wrong UDP
> checksum will be sent on the wire.

I agree with Ron.
Furthermore: one of the main goals of the current draft is "a generic method is 
defined in this document that does not require deep packet inspection at line 
rate".
The suggestion to have the tunnel termination point recompute the UDP checksum 
requires complex line rate processing at the egress LER, which may be a TC 
itslef.
I believe an incremental checksum update at every LSR is far simpler than 
recomputing the checksum at the LER.

A fair alternative is to update rfc2460, and to allow UDP checksum=0 (i.e., no 
checksum) for IPv6, very similar to UDP over IPv4.
Of course this goes beyond the scope of TICTOC, but is probably the best 
alternative.

Regards,
Tal.

________________________________
From: [email protected] [[email protected]] On Behalf Of Ron Cohen 
[[email protected]]
Sent: Friday, January 28, 2011 11:25 AM
To: Jack Kohn
Cc: [email protected]
Subject: Re: [TICTOC] UDP checksum

Hi Jack,

I don't think it is a good idea. In my opinion we should not / can not specify 
that packets with wrong UDP checksum will be sent on the wire.

Best,
Ron

On Fri, Jan 28, 2011 at 4:28 AM, Jack Kohn 
<[email protected]<mailto:[email protected]>> wrote:
On Thu, Jan 27, 2011 at 6:24 AM, Bhatia, Manav (Manav)
<[email protected]<mailto:[email protected]>> wrote:
> Yakov,
>
>> Actually, I think Stewart was referring to
>> http://tools.ietf.org/html/draft-ietf-6lowpan-hc-13 .
>>
>> If there are several proposals on the table, albeit for special cases,
>> to make the UDP checksum optional, then we can propose that
>> operation too.
>
> Alternatively, we could just keep this within our WG and add the following to 
> the 1588oMPLS draft:
>
> LSRs that adjust the CF field MAY not update the UDP checksum in case of 
> IPv6. The node that terminates the tunnel MUST recompute the UDP checksum 
> after MPLS decapsulation and before any further processing.

I would prefer this over a proposal for deprecating UDP checksum for IPv6.

Each 1588aware LSR that updates the CF is anyways updating the
ethernet FCS, which should be good enough to verify the basic
integrity of the packet.

My 0.02$

Jack
_______________________________________________
TICTOC mailing list
[email protected]<mailto:[email protected]>
https://www.ietf.org/mailman/listinfo/tictoc

_______________________________________________
TICTOC mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/tictoc

Reply via email to