Re: [Asrg] Spam, defined, and permissions

[Devdas Bhagat <devdas at dvb.homelinux.org>]

On 28/12/04 09:46 -0500, Hannigan, Martin wrote:
<snip>
> > And exactly how does the telco propose to enforce my usage of their
> > server? Post 25 blocks? I just use another port, or I VPN.
> 
> We could play this circle game all day. At some point you're
> going port 25. If you VPN to someone elses system, fine, but they
> end up paying for your traffic if you break their cap.

My point is that I control both ends of the VPN, and I have no caps at
either end.

> > > Why are you asking me these questions that you know the answers to 
> > > already?
> > > 
> > > > 
> > > > Viruses and significant volumes of spam go direct to MX, 
> > so you do not
> > > > have the luxury of forcing them through ISP controlled 
> > servers either.
> > > 
> > > Sure I do. This is not some new technology we're discussing. Is it?
> > 
> > How do you propose to do that? Port 25 blocks?
> >  
> > > > So long as you do not have centrally controlled servers 
> > for email, any
> > > > feasibility of billing for email is negligible.
> > > 
> > > I don't know a tier1 that doesn't have centrally controller servers,
> > > or any ISP for that matter. If they don't offer email services, then
> > >  I guess they don't block port 25.
> > 
> > ISPs control their servers. They do not control the servers that their
> > users can use.
> 
> They control the paths that they take. 

ISPs control layer 3 paths. ISPs do not control layer 7. Trying to
control layer 7 will merely result in SSL certificates or the like being
required, and with self signed certs, the whole ISP interception fails.
 
> [ SNIP ]
> 
> > 
> > With few entities involved, billing is relatively simple. 
> > Also, all the
> > billing is done at the exchange, not on the phone and the phone CANNOT
> > bypass the exchange.
> 
> The billing isn't done at the exchange. It's done via the SS7 network.
> The only relevance is the calling station.  
> 
> [ snip ]
> 
> > > Have you read the RFC for SIP? 
> > > 
> > > Let's use Vonage for an example. There's a call setup and teardown
> > > via a centralized server because their plans are flat rate 
> > to a measurement
> > > of minutes.
> > > 
> > 
> > However, if I want to talk to my friends in the US, I can setup a VoIP
> > network for them, or they can all install Asterix or another piece of
> > software and then talk directly. Direct end to end connectivity,
> > competely bypassing any provider.
> 
> Known ports and protocols. You could do this, but at some point you're
> going to want to talk to others. Someone is going to pay.

Sure. Anyone that I want to talk to will have their own VoIP service, or
they will hook into mine. I can become my own provider for various
services.
(This is the same reason why NAT is evil).

> 
> > 
> > You need Vonage to be able to talk to the PSTN. If you set a reverse
> > constraint that you will only talk to IP phones, that 
> > communication can
> > go direct to the other end without having to go via the providers
> > exchange.
> 
> You're describing a crippled system.

I am just limiting it to an IP only scenario, and removing the telco
from the picture.
 
> > 
> > To prevent such things from happening, you would either have 
> > to control the
> > user end device, or use proxies which would prevent such 
> > communication.
> 
> I don't know why you don't think nsp's have complete control
> over their networks. There are ways to bypass things, but you
> just end up on someone elses network. 1 to 1 isn't very useful 
> unless you're only going to talk to that one end user for the
> rest of your life.

Everyone I need to talk to currently is just using their ISP for L3
connectivity. Nothing more.

> > 
> > > > > drives entities to act by either a.de-zombify or b. not sending
> > > > > junk mail - because it's not cost effective ie. comcast 
> > > > charges a commercial
> > > > > rate for the line, a fee over the cap, a higher cost plan, etc. 
> > > > 
> > > > It would be easier to just charge users by the byte.
> > > 
> > > No it wouldn't. You'd have to measure every email instead of a start
> > > and stop record like RADIUS AAA.
> > 
> > I wasn't restricting myself to email. All your network 
> > traffic would be
> > charged by the byte, regardless of content.
> 
> Which is part of the problem. If it were broken out by protocol and
> had true costs reflected, realistic maximums could be established.
> Even breakout by primary protocols i.e. udp, tcp, smtp, rtp, etc.
> would be useful for product design. Not for end user billing. But 
> the product pricing is most important anyhow.
> 
> [ snip ]
>  
> > > Ok, flat rate. But that's what I'm talking about. A flat rate 
> > > network that charges upon violation of the SLA you pay for i.e.
> > > you pay 9.99 for 100 bi-directional emails a month, and it goes
> > > up from there.
> > 
> > But if I use their network only for layer 3 connectivity and not layer
> > 7, how do you propose that such a charge be enforced?
> 
> You're missing the point. I don't need by user granularity. I need
> a inter-carrier billing system (here today) that reflects the actual
> usage by my peer. The end user is their carriers responsibility. The
> vonage 500 minute plan remains my example.

But on the Internet, any host with a routed IP address is your peer.
This includes the end users device(s), and not only those of the NSP.
Your whole plan has a fatal flaw in this point (by design of the IP
network).

Devdas Bhagat

_______________________________________________
Asrg mailing list
Asrg at ietf.org
https://www1.ietf.org/mailman/listinfo/asrg