[saag] Re: Relative OIDs are the simpler form of OID (Re : A simpler form of OID)

On 9/29/2025 1:19 PM, Phillip Hallam-Baker wrote:
> On Mon, Sep 29, 2025 at 1:54 PM Salz, Rich <[email protected]> wrote:
>
>>     - Do we really have to argue over that to agree on my original point
>>     that the original motivation for layering every application protocol over
>>     HTTP (because it is the only thing allowed through the firewalls) has been
>>     rendered obsolete because QUIC with cheap TLS certs achieves the same goal?
>>
>>
>> If that was your original point, I missed it.
>>
> It is there in the thread.
>
>
>
>> But QUIC connectivity, as measured by HTTP3 deployment and seems to be
>> comparable to IPv6 deployment.
>>
> I don't see IPv6 as being relevant. Deployment of IPv6 is much harder for a
> whole host of reasons.
>
> Seems likely to me that new Web Services are going to shift from JSON over
> HTTPS to JSON with some framing over QUIC even faster than JSON
> displaced XML. Can't see why I would start with anything else for new work.
> QUIC is just a much more satisfactory approach.
>
> All that HTTP gives me when building a Web Service is firewall bypass,
> transaction framing and the ability to share a large number of services off
> one port and it limits me to simple request/response patterns. With QUIC I
> can easily set up a 'subscribe to stream' pattern.

I suppose that Rich was comparing the deployment rate of IPv6 and QUIC, 
but these are indeed tow very different problems. The recent thread on 
the QUIC mailing list showed that the real issue for QUIC was deployment 
by web sites, gated by administrators who could upgrade but don't 
because they won't touch a config that works, and to a lesser extend by 
web site that use server software (Apache) that does not support QUIC 
yet. The situation becomes radically different when the service deploys 
both the servers and the clients. Meta is doing that, and to quote Matt 
Joras, "For our services, where the majority (but not all) have QUIC 
enabled, we see HTTP/3's share of HTTP egress just shy of 90% globally, 
with many ASs where it is nearly 100%." So there is hope.

However, while many firewalls let QUIC through on UDP port 443 when the 
ALPN is set to "h3", it is not clear that all would allow other ports 
than 443, or other applications than HTTP3.

-- Christian Huitema

_______________________________________________
saag mailing list -- [email protected]
To unsubscribe send an email to [email protected]