[arin-ppml] An article of interest to the community....

Owen DeLong owen at delong.com
Fri Sep 2 04:33:30 EDT 2011

On Sep 2, 2011, at 12:18 AM, Matthew Kaufman wrote:

> On 9/1/11 11:41 PM, Owen DeLong wrote:
>> However, you are a corner case. A microcosm.
> I'm sorry, but "accessing web sites from browsers and web services from applications" isn't "a corner case".

No, but, an environment where that's ALL your users expect from "Internet Access Service" is.

> The fact is that almost everyone with an Internet connection at their home or business would be perfectly happy with ports 80 and 443 working well and nothing else. Perhaps the occasional random UDP for VoIP (to a hosted VoIP service that relays all traffic anyway) but even that isn't strictly necessary if your TCP 80 and 443 are working well. Even the POP/IMAP and SMTP ports aren't needed any more, as web-based mail clients are all the rage.

That really isn't the case according to most of the access providers I talk to. I suspect they know better than you.

It also isn't the case according to flow stats I have access to which seem to back up what the access providers are saying.

IM, VOIP, and gaming have all become quite popular, not to mention VPNs to reach the office network. There are many other things going on as well.

In terms of your definition of VoIP, no, that's also not how most of them actually work today except as a last resort. They may appear to work that way in your environment because perhaps you've broken all other mechanisms, but, most VOIP systems today use a direct end-to-end connection using a port-mapping protocol that allows each to establish an outbound flow through their firewall and then connect those two flows together using one or more of STUN, UPnP, NATPMP and possibly some ALG capabilities.

Lots of people (myself included) actually still use SMTP and IMAP. A few still use POP, too.

Your perception of the internet at large simply isn't in line with most user expectations. Therefore, I consider your perception of user requirements to be a corner case.

> Would this be true if there hadn't ever been NAT? We'll never know... but it the architecture in use today, and NAT and strict firewalls don't break it at all.

It already isn't true in spite of NAT. True NAT breaks many things that people used to use and has probably prevented new applications from seeing the light of day, but, nonetheless, the internet is more than just WWW.


More information about the ARIN-PPML mailing list