[Int-area] L2TP sequencing: Commonly disabled for IP data? Or always?

[Bob Briscoe <[e-mail]>]

Int-area list,

I'm looking for experience on common L2TP practice, most likely from 
operators. I tried sending this query to the [e-mail] list, as 
advised by Carlos Pignataro, but apparently it no longer exists. So I 
think int-area is the "list of last resort" for this.

The L2TP RFC says sequencing /can/ be disabled for IP data, but it 
doesn't say SHOULD or MUST. Is it possible that some operators enable 
L2TP sequencing for IP data? And if so, do you know why they would? 
Also, are you aware of any other types of tunnel that might try to keep 
IP data packets in sequence?

My reason for asking:
We (in tsvwg) are working on active queue management technology. Certain 
AQM schemes (e.g. FQ-CoDel, L4S) give lower delay to a subset of 
traffic. If the bottleneck queue supports such an AQM and it is within 
an L2TP tunnel with sequencing enabled, the egress would hold back all 
the nice low delay packets until it can put them back into order with 
the higher delay traffic.

We intend to advise that operators MUST disable L2TP sequencing if they 
wish to deploy these AQMs within an L2TP tunnel. So we need to know:

 1. Whether this will create a dilemma for any operators who need L2TP
    sequencing of IP data for some reason;
 2. Or whether we even need to bother giving the advice, because no
    operator would ever enable L2TP sequencing of IP data anyway.


Obviously, some operators already use existing technologies like 
Diffserv to reduce delay for a subset of IP data traffic, so I assume 
they always disable L2TP sequencing anyway.

Regards


Bob Briscoe


-- 
________________________________________________________________
Bob Briscoehttp://bobbriscoe.net/