Re: [Spud] [Stackevo-discuss] Fwd: Possible WG-forming follow-on to SPUD BoF

[Tom Herbert <[e-mail]>]

On Mon, May 23, 2016 at 7:33 AM, Mirja Kühlewind
<[e-mail]> wrote:
> I don’t think we can deny the existence of stateful network work devices anymore. Even though your load balancer does not work (well), there are other things such as firewall which need to keep state for a good reason.
>
Mirja,

There was a great presentation in Buenos Aires on UDP usability
(https://www.ietf.org/proceedings/95/slides/slides-95-maprg-3.pdf).
Unless I'm reading it incorrectly, the conclusion is that UDP is
mostly usable on the Internet as is, stateful devices can already deal
with it even though UDP has no concept of explicit signaling for
connection creation/tear down. Also, judging from the comments from
network operators that block UDP, it seems they are more inclined to
assume all UDP is bad and I don't think we've seen any indication that
adding explicit state signaling would change that view.

I suppose it's a natural inclination to try to mimic TCP semantics
because of its pervasiveness, but if things already work well enough
then I don't see what problem it solves. The addition of such
signaling increases the complexity of the protocols, the networking
stack, network devices, and creates yet another opportunity for
ossification. So, yes, we can acknowledge firewalls and other stateful
devices do exist, but if existing protocols already work I'm not
seeing why we have to do anything differently.

Tom

> Mirja
>
>
>> Am 23.05.2016 um 15:44 schrieb Brian Trammell <[e-mail]>:
>>
>>> I'm not sure what "path layer" is. Yes every packet takes some path,
>>> but the Internet is packet-switched not circuit-switched, so not all
>>> packets for a flow are required to always take the same path.
>>> Maintaining flow state in intermediate nodes is only best effort. In
>>> the presence of multi-homing and mobility intermediate devices that
>>> track flow state will inevitably have it wrong. Anyone who has ever
>>> tried to build a front-end L4 load balancer understands this problem
>>> all too well :-)
>>
>> Your L4 load balancer is a path layer device. The fact that the architecture (and operational practice built around it) don't work well with it is a consequence of the architecture ignoring the existence of path layer devices.
>