Email a colleague    

August 2014

The Inner Clockworks of IP eXchange: Registries, Hubbing & Policy

The Inner Clockworks of IP eXchange: Registries, Hubbing & Policy

Large enterprises around the globe spend huge sums to maintain private networks for their voice, data, and multimedia traffic.  And helping them build and enhance these networks are the Tier 1 carriers who dedicate great engineering and sales energy in that effort.  In fact, the business is accelerating because so much commerce is moving to the cloud.

But what if medium and small businesses could gain the same security, QoS, and control advantages of these private networks — and do so on an on-demand basis?

Well, that’s the vision of IP eXchange (IPX), a telecom industry movement whose aim is to commercially enable and massively expand the reach of High Definition (HD) voice, HD video, HD conferencing, unified communications and many other high quality IP services.

The GSM Association has given the IPX movement a much-appreciated publicity boost.  But the real grunt work of winning supporters and developing the infrastructure to enable IPX is being done by several vendor organizations.

One of those key IPX players is London-based XConnect, a firm with offices and facilities in the US, EMEA and Asia and whose customer base spans 70+ countries and includes major carriers, service providers, Unified Communications, and Video Conferencing providers.

I ran into Cliff Radziewicz, XConnect’s Vice President of North America, at the ITW conference in Chicago where he gave me an excellent brief on the business dynamics and technical details of IP eXchange.  He’s kindly agreed to share highlights of our discussion with Telexchange readers.

Dan Baker: Cliff, I note that XConnect has been in the IPX business for nine years now, however the city of Rome wasn‘t built in a day either.

Cliff Radziewicz: Yes, Dan.  We certainly wish IP eXchange had gone mainstream many years ago!  But at least our vision has remained the same those 9 years: to help interconnect everyone to advanced IP services in voice, video, UC and many others using a policy-rich approach.

Today we’re very optimistic because we think the time gap between vision and implementation is narrowing and XConnect has the experience to capitalize on this trend.

So if I may, let me walk you and your readers through a quick explanation of how IP interconnect works on a business and technical level.  Along the way, I’ll explain XConnect’s business model.

The Technical & Business Mechanisms
behind IP eXchange

The Need for Guaranteed End-to-End IP Delivery

Perhaps the best way to understand what end-to-end connection of IP services is about is to consider the old TDM world.  In that circuit-based world, calls were routed based on the phone number prefix.  So country code 44 connects you to someone in the UK.

But today, if you dial 44 on a mobile phone, it might end up in New Jersey, because the numbers are not necessarily tied to geographic-based routing any more.

However, even a VoIP-originated call doesn‘t have to be routed end-to-end via IP.  It can start IP, go to TDM, get converted back to IP and so forth.  But video and other IP multimedia doesn’t permit that: you need to guarantee a full IP path end-to-end, so much greater guidance and control of the traffic is necessary.

IP eXchange Example

Querying a Registry Database for Answers to Routing Questions

To complete an IP exchange, the network needs many questions answered.  For instance, we need to know what kind of call Carrier A is making.  Is it a video call, a HD voice call, etc?

And what route should it take?  Is Carrier A allowed to route a call to Carrier B?  Will Carrier B even accept a call from Carrier A?  Carrier B may also decide that it only wants to accept video IP calls at such and such IP address.  And maybe High Definition Voice goes to another IP address and they accept standard IP voice calls at still another.

The TDM network has no clue about all this, so this is where a “registry” database in the cloud is needed to enable operators to get all the intelligence they need about making IP connections.  And XConnect provides such a service.

Carriers query our registry using ENUM or SIP protocols.  The intelligence they get in return answers questions like: Does the number I want to call actually exist?  Who does it belong to?  How do I get there?  Does the terminating number support the requested service?  Then, based on the data the registry provides, Carrier A makes a decision on what to do with the call.

So this is the purpose of the registry and an originating carrier will basically pay for each database dip.

Example: An HD Call from Orange France to Verizon USA

OK, let’s walk through a hypothetical example of how an end-to-end HD voice interconnect call is made.  We’ll use Orange in France as the originating carrier because they’re one of very few operators in the world who have enabled HD voice across their network.

Let’s suppose that Orange wants to complete its HD voice call to Verizon in the U.S.  And let’s say that HD voice is enabled for 20% of Verizon’s numbers.  I’m making these figures up.

Assuming that Orange subscribed to XConnect Registry Services, the first thing Orange does is initiate a query to the XConnect registry asking how to connect to telephone number: +1-201-777-9999.  Well the registry responds saying, “Route the call to this IP address, and that phone number is HD enabled” -- and maybe provides additional information related to the called number.  So the registry flexibly manages the many parameters of connection.

So Orange now has the vital data it needs to complete the call.  It knows the call is going to Verizon and the number can accept HD.  The next question Orange has is: Do I have a direct IP end-to-end connection with Verizon?  And if they do, they route it that way and Verizon completes the call.

And if there’s no route for HD, then Orange will send a standard bandwidth call, not an HD one.

IP Hubbing: XConnect Completes Orange’s HD Call

Now, let’s further suppose that Orange does not have a direct IP end-to-end connection to Verizon.  Well, in that case, Orange may decide to let XConnect connect the call for them.  So here, XConnect becomes the actual interconnect service provider, charging Orange to complete the call.

We are completely agnostic about which path Orange chooses.  If a registry dip is all they need, that’s fine.  XConnect’s philosophy is to give the operator the flexibility to make their own best economic and service decisions.

In the larger scheme, XConnect competes with other registry providers such as Neustar and iconectiv.  We are not the only IPX hubbing provider out there either.  There’s Telx, Equinix and many others too.

Frankly, a great deal of coopetition occurs between these interconnect players.  For instance, sometimes we offer our IP hubbing or registry services to our rivals who maybe need help reaching some corner of the world where we have better connections to regional players or where we provide better access to data.

The Role of Co-Location and Fiber Interconnect Players

Up to now, I’ve only talked about the Layer 3 and above intelligence required to make IP interconnect work.  Getting the actual physical connection often requires working with firms who specialize in telecom co-location and fiber connect facilities.  So companies like Telx and Equinix provide those Layer 2 connections with local providers in various regions of the world.

Our approach to working with these firms is to form a federation — basically a community of interest among service providers in a particular geographic location or those focused on a particular service such as video or HD voice.

For instance, we just announced a federation in Malaysia.  We also have working federations in South Korea, South Africa, the Netherlands, Norway, and Germany.

So in these federations we partner with the equivalent of a Telx a co-location provider.  And that partner will sell our services to all the local carriers in the country.  So the colo providers sell the physical connection and XConnect provides the Layer 3 and above services including the registry services, session border controller, the softswitch, the voice termination, customized routing solutions and local number portability database data.

Thank you for this very nice summary of IP eXchange.  And let me further ask about the standardization of IPX.  Will we see one or several flavors of IPX emerging?

While it’s convenient to talk about IPX as if it were a global standard, its implementation will widely vary because service providers and enterprises want to customize and won‘t be satisfied with something plain vanilla.

And this is where XConnect aims to differentiate by enabling a policy-rich IPX.

For example, there are competitive reasons why carriers want a more secure means of IP exchange.  In South Africa we helped form a federation of 40+ operators and some of those operators are fierce competitors.  So a policy-based exchange capability is very useful here to allow Carrier A to say: “Everybody in the South Africa Federation can complete video or HD calls to me, but not Carrier B.”

Likewise, the commercial enterprises who will drive a big share of IPX demand want better control of their communications.

In many cases, enterprise customers require the same flexibility, functionality and security as telecom service providers, so this is where XConnect’s policy-driven registry services and SIP interconnect services allow those enterprises to create customer, supply-chain and related voice/video federations and customized routing solutions on their own terms.

Cliff, I guess a lot of readers would be very curious to ask: what’s it going to take for IPX to go mainstream?

Dan, we’ve conducted tests of federated HD voice in the U.S. and UK markets with our partners and the service works fine.  We’re all waiting for the larger mobile or fixed service providers to embrace HD voice and offer handsets and solutions that accommodate HD voice on a generally available basis.

As I mentioned, Orange in France has made that commitment.  I’m sure there are other service providers implementing HD -- or considering implementing HD -- right now.  But unless many other network providers can easily interconnect with Orange and with each other to offer end-to-end HD voice calling, the value of the HD voice service to the consumers within an individual HD ‘island’ is generally limited.

So until a few more of the larger players offer HD as part of their product set, it’s not going to skyrocket like we fully expect it’s going to some day.

When the market does take off, we think our approach of offering service providers the utmost in flexibility is going to prevail.  And that’s a function of how well you implement the policy aspects of the service.  It’s policy that enables carriers to choose at what commercial, security and QoS levels they need to interoperate.

Copyright 2014 Telexchange Journal


About the Expert

Cliff Radziewicz

Cliff Radziewicz

Cliff Radziewicz is currently Vice President — North America for XConnect.  He joined XConnect in 2010 from Arbinet Corporation, where he served as Vice President of Business Development.  At Arbinet, Cliff was instrumental in evangelizing the benefits of IP-based voice peering and federation business models to telecom carriers worldwide.

Cliff’s 25 years of telecom experience includes serving as a corporate executive at Bell Atlantic and AT&T, as well as an entrepreneur and founder/CEO of several private equity funded telecom and e-commerce businesses.   Contact Cliff via

Related Stories

Related Articles