Re: [netconf] draft-ietf-netconf-udp-notif-10

On Wed, Aug 9, 2023 at 12:41 AM Jürgen Schönwälder
<jschoenwaelder@constructor.university> wrote:

> On Mon, Aug 07, 2023 at 05:19:17PM +0200, Alex Huang Feng wrote:
>
> > Something we realised during our tests is that in Linux, when the packet
> passes the MTU and it fragments at IP level, the performance drops greatly
> (in terms of throughput and amount of messages we can deal with).
>
>

Agreed.
We used to set the chunk size to 2000 many years ago.
Things worked better changing it to 1000.

The key section here is 4.1
https://www.ietf.org/archive/id/draft-ietf-netconf-udp-notif-10.html#name-segmentation-option

*Consequently, implementations of the mechanism SHOULD provide a
configurable max-segment-size option to control the maximum size of a
payload.*
Each segment is the same as a chunk in NETCONF or RESTCONF message framing.
A server would not actually use the maximum size of 65535 if the MTU is
1500.

> You more or less repeat the IPv4 fragmentation design above the UDP
> layer and it seems by doing so you inherit all its problems. It feels
> like you are moving the problems around rather than solving them (and
> if you go there you might end up with something that you do not want
> in your aim to be fully stateless).
>

I agree this is a concern.
The chunking feature of NETCONF or HTTP is easy to use over TCP where none
of this extra work is required.

> Perhaps you just wanted UDP and the rest just got added for political
> correctness in the IETF?
>

The solution is partial at best and designed for use in enterprise networks.
The introduction describes lots of line cards blasting UDP packets at a
single receiver.

Maybe it is unwise to have a standard that almost certainly generates
packet loss,
yet has zero tolerance for packet loss.

> /js
>

Andy

>
> --
> Jürgen Schönwälder              Constructor University Bremen gGmbH
> Phone: +49 421 200 3587         Campus Ring 1 | 28759 Bremen | Germany
> Fax:   +49 421 200 3103         <https://constructor.university/>
>
> _______________________________________________
> netconf mailing list
> netconf@ietf.org
> https://www.ietf.org/mailman/listinfo/netconf
>