Easier thing we can do, is to map them down like we do it with IPs and domain names, but as usual, this is far from being a practical solution.
Best think is to let conventional telephony die and VoIP take over its place.
We have been able to map phone numbers to SIP URIs for a long time with E.164 ENUM. Numbers are a hassle, but there's nothing preventing numbers from living alongside proper URIs today.
> let conventional telephony die and VoIP take over its place.
Many places are still circuit-switched at the last mile, whether over landline or cellular. We can't turn off the PSTN in favor of VoIP until everybody has switched over -- similar to the IPv6 changeover dilemma, although more tractable due to the ubiquity of Internet access and proper interworking -- no tunneling required.
What I'm waiting for though, is the ability to "bring your own" SIP URI so the carriers can handle our calls in the same way GMail can handle our email for personal/business domains. (Not to mention the capability to dial out to SIP addresses...)
Ideally, this would extend all the way down to "bring your own SIP proxy and registrar" where the carriers would just act as dumb pipes to carry your calls at the same QoS as theirs, but that's a far ways off.
yes, because dialing 3.0.3.3.8.4.0.3.5.5.1.2.5.e164.arpa is super easy to do.
I was talking about a proper solution, not patching an existing one.
> We can't turn off the PSTN in favor of VoIP until everybody has switched over
Exactly, but we have to start at some point. We have been doing phone calls for more than 100 years, and it remained practically unchanged since then.
Nonsense, the user would never dial that. The number gets mechanically translated as necessary, which is extremely straightforward...
> We have been doing phone calls for more than 100 years, and it remained practically unchanged since then.
Really? Most of your comments in this thread are uninformed, bordering on the absurd. If you are doing a telecom startup (as your profile implies), I would encourage a better understanding of the industry: you can't compete with what you don't know.
(/s)
That's because circuit switching is literally setting up a wired connection between two points.
Good luck trying to be that reliable with a population the size of the Internet, and if you can, good luck being able to pay and maintain an infrastructure of that size.
> T.38 and RFC-2833 exist
Really? The last time I sent a fax was 2006.
You're clearly missing the point here. T.38 and RFC-2833 largely exist because there's many, many voip routes that can't transport low latency audio reliably. There's a lot of other things you can transport in a phone call that don't have out of band standards.
If you require out of band standards to reliably transmit a fax or even something as basic as DTMF, you're clearly stifling what can be done within a perfectly valid form of transport.
> Good luck trying to be that reliable with a population the size of the Internet, and if you can, good luck being able to pay and maintain an infrastructure of that size.
Several companies have, and continue to do just that not just around the country, but around the world; end offices that serve customer lines are connected together by trunks to tandem switches. You clearly have no idea how the phone network works.
> That's because circuit switching is literally setting up a wired connection between two points.
It's not just one channel, though; techniques such as time or frequency division multiplexing allow you to stick thousands of calls over a cable or RF link.
Most VoIP calls are also carried as plain data traffic. Digital voice services, which are basically VoIP (VoLTE, AT&T U-verse, etc), properly differentiate between calls and data, so calls on these networks are going to be much more reliable than simply leasing a SIP trunk. Once reliability is in place, audio quality is certainly going to be better between compatible endpoints.
You say that as if things like GR-303 and ISDN BRIs/PRIs don't exist. The only thing that isn't digital about modern, non-IP phone networks is the last mile when you order an analog loop.
> properly differentiate between calls and data, so calls on these networks are going to be much more reliable than simply leasing a SIP trunk.
In theory, sure. As I said though, in my experience the vast majority of these routes tend to fall behind their circuit switched counterparts in performance even on carrier grade routes.
Granted, "digital voice" was imprecise terminology, but you get the gist... modern consumer phone services are VoIP.
> fall behind their circuit switched counterparts in performance even on carrier grade routes.
Most carrier backbones have been all-IP for years. Pretty sure end-to-end circuit switched calls are atypical nowadays, although I'm happy to be proven wrong.
That certainly depends on what network and where. If you're talking about someone like Level 3, they definitely use IP transport for phone calls. But there's a number of networks that don't. Verizon, for example, has a toll network based on the DMS-250 and DEX-600 switch platforms. Both of these are designed from the ground up to handle circuit switched traffic. There are a few routes that're IP, but they're pretty easy to pick out since there's a combination of comfort noise, significant latency and packet loss (!) on them.
AT&T, similarly has a toll network running on 4ESS, 5ESS and DMS-250s. There are a few IP trunks on there, but they're typically only used for destinations like San Francisco to Chicago.
Sprint has an all DMS-250 network at their disposal as well, as well as TDS Metrocom and the McleodUSA network that Windstream bought. Anyway, you get the idea.
On the local end, to get transport to the phone network, you have to order DSX trunks; there is no IP trunking to be had. Not to mention, there's typically quite a large number of circuit switches in any given market. Typically enough to outnumber the number of softswitches.
No one is saying VoIP is as reliable as PSTN. Not even mobile (GSM, CDMA, etc.) is.
Not only do you not have to deal with the problems of internet routing, you aren't stuck with underwater sounding CELP codecs.
Anyway, if you're arguing voip should replace the PSTN, I'd hope you'd at least have something hinting it could be as reliable.
I did, and the reason some calls sound horrible, is because Skype is trying to save up some money hiring shitty carriers/providers.
> it's relatively easy to use a B channel to transport g.722
Why would you use an HD codec to transcode to G711? Call on any direction can sound only as good as G711.
> Anyway, if you're arguing voip should replace the PSTN, I'd hope you'd at least have something hinting it could be as reliable.
I agree with most of your arguments. VoIP isn't ready now, but it will be soon. Just remember how the Internet was 10 years ago, and how it is today.
The idea is to use the timeslot g.711 normally would sit in to transport your own codec. Studios use this technique pretty frequently when they hire voiceover talent or do remotes.