Re: [mpls] Last Call: <draft-ietf-mpls-in-udp-04.txt> (Encapsulating MPLS in UDP) to Proposed Standard

[Gregory Mirsky <[e-mail]>]

Hi Stewart, et. al,
I think that the method to detect degradation of a tunnel service can use any of performance measurement metrics, e.g. packet loss, delay or delay variation. An operator may have controls to define what constitutes entering and exiting Severe Service Degradation error state, e.g. threshold, number of consecutive down threshold and up threshold measurements, etc. And associate actions with entering and exiting error state.

	Regards,
		Greg

-----Original Message-----
From: mpls [mailto:[e-mail]] On Behalf Of Stewart Bryant
Sent: Wednesday, January 22, 2014 2:09 AM
To: Eggert, Lars; [e-mail]
Cc: Joel Jaeggli; [e-mail]
Subject: Re: [mpls] Last Call: <draft-ietf-mpls-in-udp-04.txt> (Encapsulating MPLS in UDP) to Proposed Standard

On 22/01/2014 07:51, Eggert, Lars wrote:
> This is not at all the argument I am making. My emails have not touched at all on the issue of zero checksums.
>
> My point is that UDP encapsulation changes the potential *reach* of congestion-uncontrolled traffic that was otherwise limited to L2 networks.
>
> Lars
How about if text is introduced recommending the use of an OAM between tunnel endpoints that monitors packet loss.

The tunnel endpoints need to know if the tunnel is broken anyway and on hitting a loss threshold they can  alarm, redirect the traffic, or shutdown, depending on configuration, topology and the needs of the operator.

You probably also need pro-active CV to make the whole system work.

This many not necessarily be fast or elegant, but if there is significant misdelivery or congestion loss, traffic will eventually cease.

Stewart
_______________________________________________
mpls mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/mpls