Transport Protocol Adaptation (Re: [RAM] Re: the separation of ID/RLOC)

[Christian Vogt <christian.vogt at nomadiclab.com>]

Iljitsch van Beijnum wrote:
>> One possible improvement would call for an edge network operator to
>>  redirect part of its traffic to an alternative provider as the 
>> delays incurred by the current primary provider increase.
> 
> This is susceptible to oscillations.
> 
> In my opinion, the correct way to handle this stuff is with multipath
>  transport protocols.

Having worked on mobility management for several years, I fully agree
that, for efficiency reasons, the transport layer must be made aware of
connectivity changes at lower layers.  This holds for both mobility and
multi-homing.

Purely relying on multi-homing support in transport protocols, however,
would suffer the same disadvantages as Shim6.  It would limit
edge network operators in their freedom to pursue traffic engineering.

Purely network-based solutions, on the other hand, make it much harder
for a transport protocol to adapt to provider changes, because they do
not receive explicit indications of such changes.  (IP addresses are
provider-independent, after all.)

Only combined host/network-based solutions can bridge the gap.  They
allow edge network providers to do what they desire, and at the same
time provide the host with timely information on provider changes, which
is what the host's transport layer needs to adapt.

> With IP, "backup" paths are rarely idle when the "primary" path is 
> active, so loss of the path that's currently used means a reduction 
> of total available bandwidth which can easily reduce this to less 
> than the amount of traffic generated.

Ok.

> So backing off until a new equilibrium is found is a good idea.

Without question.

>> if the mapping function would allow you to do this quickly enough 
>> to not impact ongoing communications too much.
> 
> If you were using a TCP-like transport protocol that works over 
> multiple paths at the same time, you'd already have RTT times and 
> window sizes for the other path so switching could happen without 
> reducing bandwidth more than necessary. And load balancing would 
> happen automatically.

Yes, I agree.

Take care,
- Christian



_______________________________________________
RAM mailing list
RAM at iab.org
https://www1.ietf.org/mailman/listinfo/ram