dslreports logo
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
uniqs
1038
share rss forum feed


ThaiGuy

join:2008-05-10
Thailand

RTP media on VoIP calls to PSTN

I have a Cisco ATA in Australia making calls to an Aussie PSTN number via CallCentric. Everything works pretty well but the RTP goes through the CallCentric proxy. I made sure to use STUN and set the codec to G711a in the hope that RTP would be sent directly to the PSTN termination provider in Austrailia (whoever that may be) but no luck.

I have not tested VoIP.ms yet but I am looking for a provider that would be able to help me overcome this problem.

I read that CallWithUs and Flowroute may allow it. Can anyone confirm this or point me in another direction?

PX Eliezer70
Premium
join:2008-08-09
Hutt River
kudos:13

1 recommendation

From the owner of CWU a few years ago:

[CallWithUs server auto detects NATed SIP clients. If the client is on a public IP address (or STUN is working properly), then direct audio path between call termination GW and SIP client is set up. Otherwise audio is proxied by one of CWU servers closest to SIP client geographical location (identified by SIP client external IP address).]

In addition to CWU, FlowRoute and Voipo do similar AFAIK.


ThaiGuy

join:2008-05-10
Thailand
Thanks. Which of the 3 providers would be best suited for me to manage a large number of accounts?

PX Eliezer70
Premium
join:2008-08-09
Hutt River
kudos:13
said by ThaiGuy:

Thanks. Which of the 3 providers would be best suited for me to manage a large number of accounts?

I'm no expert on that....

CWU does allow multiple registrations/subaccounts on one account....

FlowRoute seems to offer many business-friendly services.

Voipo does not really support BYOD very much these days.

Iscream
Premium
join:2009-02-17
New York, NY
kudos:6
Reviews:
·Verizon FiOS
reply to ThaiGuy
I still can't understand what is exactly your "problem" you mentioned in original post?

It doesn't matter _how_ your media goes - what matters is the voice quality (your own and your called party's subjective feeling about the call, measured in delays [latency] and/or voice breaking-in/out, etc.) you get on both ends of a call.

Do you experience voice quality issues?

Callcentric, as well as most likely any other VoIP provider, has multiple carriers who can terminate your call. Those carriers are either larger or smaller aggregators/resellers or some national/international/local carriers.

A carrier (their switch) used to terminate any particular call can be in any part of Earth. Just any. The carrier may interconnect via TDM (PSTN line) or via IP protocol (not necessarily SIP). It can be BT (British telecom, just to name one) division in Brazil or it can be TATA whose switches are in US, Europe, India (mainly), etc. Therefore your call can be terminated by any [geographically not relevant to your origination point] _currently_ used for your dialed route/destination switch or operator.

Also, as it was admitted by PX Eliezer - there is a NAT issue, most (99.99%) VoIP users are normally behind NAT; on another hand - most Tier-1 carriers NEVER keep their networks open for 3-rd party RTP - pure security measure.


ThaiGuy

join:2008-05-10
Thailand
If the client ATA is in Australia and the carriers switch is in Australia, my problem is the potential for introducing latency by sending RTP packets to the USA only to have them relayed back to the carrier in Australia.

Iscream
Premium
join:2009-02-17
New York, NY
kudos:6
Reviews:
·Verizon FiOS

2 edits
Yes, that's the case, but again - you shall worry about an actual voice quality, stability and billing (based on proper signaling), not by a theoretical latency.

If that's an "obsession" point (I can understand you there), because you think "local" calling versus "global" (although you never know how your packets travel while it's possible [although unlikely] for your "domestic" packets to travel via Tokyo in Japan or any other regional or even far away backbone ) - then I'd suggest using at least two VoIP providers (you can use many more) whereas one provider will be used for local calling only (I believe you can find a dozen, if not more, providers in Australia) - you can program your user devices to route calls based on digits dialed - this is called a "dialing plan" or dial-plan; and setting all international calls to be sent to your other provider (or providers).

You can set multiple providers for your international calling - also depending on digits dialed.

But really, speaking of Callcentric (because I work there) - we have quite large number of happy Australian customers (as well as in general - a substantial portion of our customer base is far away from America) because Callcentric is a regional ISP in NYC thus having direct backbone links (called BGP4 peering) with many international ISPs thus providing minimal latencies to most places in the world.

Our average round-trip time (RTT) to Sydney is 276ms - that's a latency of 138ms (latency is measured half way of RTT) while recommended by ITU-T max. latency should be less than 250ms.
Our max. RTT during last 24 hours was 441ms (latency of 221ms), but such "bursts" are very rare.

We have a monitoring system plotting graphs to all destinations where we have direct links to; the system sends alerts whenever a RTT value is approaching a "red" zone. Then our routing/network engineers come to rescue ).

I'll try to post a graph of RTTs for past 24 hours from our switch in NYC to peering point in Sydney here - never done it in past, but saw many others doing it. Bear with me

Edit: Okay, the graph [in PDF file format] posted - above my text; fixed some typos.

If anybody interested - I can provide similar graphs to most Internet-wise large regional backbones on Earth, including historical info for last week, month and year. Just PM to me.

OZO
Premium
join:2003-01-17
kudos:2
If you work for Callcentric, why Callcentric doesn't offer its customers what they ask for - the direct media mode?

ThaiGuy See Profile certainly knows how to handle potential NAT problems and he needs and specifically asks for that option. Why do not simply help him and offer that option? Or you prefer letting him to go?
--
Keep it simple, it'll become complex by itself...

Iscream
Premium
join:2009-02-17
New York, NY
kudos:6
Reviews:
·Verizon FiOS

1 recommendation

I don't see a logical correlation between the fact that I work for Callcentric (it's very well known to residents of this forum) and you saying that "what customers ask for - the direct media route". Humans are not being logical )

Well, this matter is complex enough hence cannot be explained in a few words or sentences. You may search for it - there were lots of heated discussions of it in past.

In a very digested form and not intended for another discussion here it's again:

- most "normal" (Tier-1/2/3) national and international carriers do NOT allow media streams (RTP) to come from undeclared sources (all IP addresses must be provided during provisioning and configuration time);

- most customers are behind NAT and [despite your words] they can NOT and do NOT know how to deal with media routing - this is statistically (by millions) proven information;

- there are CALEA (legal intercept) requirements which Callcentric as being a licensed national (CLEC) and international (IXC FCC's 214 licensee) carrier must abide by;

- our own measurements (and there were also lots of discussions of this matter at this forum) show that there are NO reasonable merits to be concerned about such thing as a "latency" when network is setup and operates according to published standards. Problems rise [without regard to direct media routing] when a service provider has a lack of resources in one or another way. But then having direct media routing won't help either. Not to say that on a memory of this generation most international calls were made by a means of satellite long distance... with latencies over 1000 ms - not VoIP, I'm talking about very recent past of PSTN - still civilization survived ) I must stress it again - studies and standards tell that a latency should be below 250ms - then human brain won't notice it (I'm not talking about X-men; just about a "normal" human brain).

- most known to me established VoIP providers and carriers do NOT provide direct media routing. You may ask this forum.

- carrier equipment (switches) sold in US by official vendors - do NOT even have an option of allowing direct media routing (Genband, Lucent-Alcatel, etc).

Again - although I provided the reasons, I don't intend to continue any further discussion of this subject. I'm apologizing upfront.

OZO
Premium
join:2003-01-17
kudos:2
said by Iscream:

- most "normal" (Tier-1/2/3) national and international carriers do NOT allow media streams (RTP) to come from undeclared sources (all IP addresses must be provided during provisioning and configuration time);

Your service will not get my IP during provisioning and configuration time, because I use smart phone as a SIP client and it always has different IP (depending on my location). And AFAIK, keeping static IP for SIP client is not a requirement for most national and international carries...

- most customers are behind NAT and [despite your words] they can NOT and do NOT know how to deal with media routing - this is statistically (by millions) proven information;

My words were about specific customer, who can handle NAT well. And I'm confident, that there will be more then one BYOD customer/s, who want to use ti, if that option was offered to them.

Moreover, if you offered direct media mode option to your customers I may be start thinking to open an account with Callcentrinc too...

- there are CALEA (legal intercept) requirements which Callcentric as being a licensed national (CLEC) and international (IXC FCC's 214 licensee) carrier must abide by;

It that's true, then why other providers don't use it as an excuse? There are providers that offer direct media mode in the US. For example, CallWithUs:
said by Arne Bolen :

One of the biggest advantages is the direct media from the carrier to the user. Direct media improves the audio quality and reduces latency to a minimum. It works like a charm.

And I happened to agree with that too.

- our own measurements (and there were also lots of discussions of this matter at this forum) show that there are NO reasonable merits to be concerned about such thing as a "latency" when network is setup and operates according to published standards. Problems rise [without regard to direct media routing] when a service provider has a lack of resources in one or another way. But then having direct media routing won't help either. Not to say that on a memory of this generation most international calls were made by a means of satellite long distance... with latencies over 1000 ms - not VoIP, I'm talking about very recent past of PSTN - still civilization survived ) I must stress it again - studies and standards tell that a latency should be below 250ms - then human brain won't notice it (I'm not talking about X-men; just about a "normal" human brain).

It's a lot of words instead of helping the customer with his needs. If he needs that, he has his reasons. If Callcentric needs his money in its revenue stream - just listen to him. I think it should be simple like that in a market economy...
--
Keep it simple, it'll become complex by itself...


ThaiGuy

join:2008-05-10
Thailand
Thanks Iscream and OZO - I appreciate your help.

Some points to note: I have been using CallCentric for my Australian customers without any issues to date. That said, it's a very small number. I expect to get hundreds of new Customers over the coming month's so that's why I need to make every effort to get the best possible solution in place and hence my research.

I have setup a test account with CallWithUs and they advised "As soon as the device behaves like on a public IP address, our server sets up direct RTP path". I am due to run some tests in the next few hours and Wireshark will give me the good or bad news.

The killer line from Iscream is: most "normal" (Tier-1/2/3) national and international carriers do NOT allow media streams (RTP) to come from undeclared sources.

So, I am worried that if CWU sets up a direct RTP that it will be rejected by the Australian carrier. Let's find out.....


Trev
IP Telephony Addict
Premium
join:2009-06-29
Victoria, BC
kudos:6

1 recommendation

said by ThaiGuy:

The killer line from Iscream is: most "normal" (Tier-1/2/3) national and international carriers do NOT allow media streams (RTP) to come from undeclared sources.

So, I am worried that if CWU sets up a direct RTP that it will be rejected by the Australian carrier. Let's find out.....

Presumably, given that we don't often see people complaining about no audio with CWU, they know what they are doing and wouldn't set up direct audio with a carrier they know won't support it.
--
I represent AcroVoice, a full service Canadian VoIP Provider.
Buy your Obihai ATA shipped from within Canada.

PX Eliezer70
Premium
join:2008-08-09
Hutt River
kudos:13
reply to OZO
said by OZO:

I think it should be simple like that in a market economy...

It's not so simple, and it's not a fully market economy when government regs are involved.

And yes, for various reasons a facilities-based provider [is] subject to more intense regulations, especially when CLEC licensing and FCC 214 licensing is involved.

You mention the example of CWU. Well, CWU is great, but they are not on the radar screen of ANY state or national regulatory authority in the US.

CC is in a TOTALLY different legal and regulatory environment from CWU, and even in a different environment from many/most of the other providers.

-------------------

When Iscream spoke of [IP addresses must be provided during provisioning and configuration time] I believe that he was talking about the setup of each specific phone call (each session). No one is talking about a static IP.

-------------------

CallCentric is a prudent company.

They would [not] be like the football team that recruits a famous quarterback then mysteriously refuses to use him.

If they think that providing a particular service to a small number of potential customers will not benefit the company overall (even perhaps result in harm) they will demur, and they will send off those fellows with a hale and hearty handshake, and their best wishes.


ThaiGuy

join:2008-05-10
Thailand

3 edits
My testing might take a little longer than I expected as I have just applied the latest Cisco firmware to my SPA112. Device registered with CWU immediately but STUN does not work. Changed to use NAT Mapping instead and now cannot register.

Ahhhhh Cisco

Modem & SPA reboot fixed the registration problem. Now I'm getting an Invalid number when I dial 611345xxxx

Progress Report: Dialed my UK VoIP number instead and call was accepted by CWU. Received a 183 Session Progress, RTP was sent direct but call didn't get connected. Will keep trying...

Iscream
Premium
join:2009-02-17
New York, NY
kudos:6
Reviews:
·Verizon FiOS
reply to ThaiGuy
As Trev wrote - CWU is trustful and dependable provider. You may rely on them.

As I wrote - due to above outlined reasons Callcentric cannot provide direct media; it's like 2x2=4.

Last thing - if you don't have _latency_issues_ today (while using Callcentric for local Australian calls) - you won't have such issues tomorrow.

I'm not saying that you won't have issues with Callcentric at all - although built to be dependable, CC is just a "normal" provider/carrier which is susceptible to all possible "normal" technical problems beyond its control, but the idea I'm trying to convey here is that the network, CC has already deployed and keeps building on a daily basis, is developed considering latency metrics hence our partner carriers participating in call origination and termination are carefully chosen and tested accordingly - to deliver said metrics.

Callcentric is in telecom (in traditional meaning of this word) business since Aug. 2001. And believe me - when TATA (just to trow out some name) sends us a call to be terminated in Hiroshima or in Paris or in Moscow or in St. Vincent - they send it to us according to their quality metrics while perfectly knowing that our switch is in NYC - they don't care about _how_ we get that call delivered. The only thing they care about - it's the quality of the route.

Because the game is very simple - if their clients can't use the route - they remove TATA from routing - therefore TATA removes "us" from routing first - in order to avoid problems with their clients.

The opposite is also true - when we send a call to TATA to be terminated in some rural Brazilian village - we don't care that their switch is in Mumbai (just to point out a name) - we only care about quality metrics.

No one asks about direct media or any other BS like that. It's that simple. Same is with us - while/when our subscribers happy - we keep respective carriers in 1/2/3 routing positions and they get steady volume of called minutes.

The moment the quality metrics change to decrease - no emotions there and no sentiments; it's automated switch's monitoring system's decision - other carriers are substituted to same 1/2/3/etc. routing positions. Quality metrics are measured in real time and calculated every 5 minutes. Routing updates are made every 4 hours while emergency substitutions may be performed at any moment.

It's called RQI - routing quality index. The measurement of RQI is very simple - ACD is multiplied by ASR multiplied by FAS ratio (ACD - average call duration. ASR - answer/seizure ratio, FAS - false answer supervision). More details are available at CC's wholesale WEB site.

Moreover - carrier world will never use direct media - because the moment we expose IPs of our carrier for a particular destination - our buyer{s} would be able to go to that carrier direct (why should they pay to a reseller). Same is true for ALL carriers. Telecom represents a full and multiple times "meshed" mesh network .

The only thing the carriers don't want to know about (with exception of fraud) - it's to know about dynamic IP addresses of their buyers' residential customers.

Such a thing would open them to security threats and made susceptible to denial of service attacks. This is why the IP addresses of each wholesale client and wholesale buyer (the roles are logical and change depending on traffic direction) are always "hard" provisioned within each carrier's network.

Yes, that's right - there are some carriers who do allow "open" RTP streams (signaling IPs are still fixed and predeclared). But to my [daily] experience - this is rather exception than rule.


ThaiGuy

join:2008-05-10
Thailand

1 edit
Thank you Iscream for sharing your knowledge on this subject. It seems like my concern about direct media is misguided and I am now confident I have made the correct decision choosing CallCentric as my main provider.

All that remains is for me to provide failover to a second provider and I will decide between VoIP.ms, CallWithUs or OnSip for that.

Update: Although I cannot make calls to Special Service Australian numbers starting with 1345, I have made successful calls to the UK and New Zealand where the RTP went direct.