Interview with Eli Katz, CEO, XConnect
Let us start with numbers. What is the number of customers you have, number of POPs, amount of traffic you are handling etc.
These are numbers that we are comfortable to release: We have over 500 million numbers in the registry. That consists of a combination of blocks of numbers allocated to service providers 
and individual numbers. We have now just over 400 service providers on a worldwide basis utilizing our services from pure ENUM registry exchange all the way up to a complete federation model whereby we are delivering not just registry services but also signalling services, security services and other wide range functions. We have millions of minutes flowing through our systems and that continues to grow.
How many ENUM queries are we talking about per month?
We have a few hundred million ENUM queries occurring on a monthly basis.
These few hundred million ENUM queries you handle exclude the in-country federations. Right?
Yes that is correct.
Out of the 400 service providers that you serve, approximately how many utilize your capabilities other than registry queries, stuff like call signalling and security?
Probably around 100+.
The 500 million figure that you gave, does it also include numbers from LNP registries?
We have only just started working on LNP integration so that the information from these LNP databases is available to the XConnect registry. There are a small number of countries that have LNP with central database services. We have developed interfaces into these databases. That will then complement our registries in terms of providing additional routing information for numbers. An important thing to mention is that as registry model continues to evolve, for a given phone number there can be multiple routes to that line. LNP database might contain some information. That may be one route. There might be some additional routing information from service provider that hosts that number. In addition if it is a PSTN call there might be information from alternative provider on a transit basis.
Out of 500 million numbers how many are VoIP end points?
It is difficult to judge precisely. But VoIP is growing in line with industry expectations. You have countries like Netherlands where upto 30% of all phone numbers and lines being delivered today are VoIP. There are also some other European countries where VoIP penetration is in the late teens and early twenties. France being one of them.
Let us just say that there are 50 million VoIP subscribers worldwide. How many of those phone numbers do you have in your ENUM registry?
A significant percentage. It is difficult to tell how many of those numbers we have in our registry. We could have numbers from service provider that is internally migrating numbers from old to new i.e. something delivered as POTS line is now being delivered as VoIP line.
Can you give a geographical breakdown of those numbers?
We don’t give breakdown by country but there is a significant percentage in the US. In Europe, UK and Netherlands are the biggest contributors.
Perhaps a 40:40:20 split geography wise with North America and EMEA accounting for 40% each and the rest of the numbers being based in APAC and CALA?
That is not a bad estimate.
In the federated model, do you get a cut for each minute or you charge a management fee?
Our usual model for in-country federations is a managed service fee.
What is the revenue model for services other than federation management?
Where the service is more like the classic settlement, the easiest model to apply is on a per-minute basis. Our pricing structure is flexible depending upon which combination or our services they use: registry, signalling, security, commercial.
Can you explain the value chain?
Registry involves provision of data, access to that data, synchronizing of various databases, policy setting etc. Signalling involves interoperability with other providers such as in a federation, sip/h323 protocol variations etc. Security: things like SPIT protection, spoofing, caller ID validation. Commercial which sits at the top of the value chain splits into two types of services. One is where we provide settlement services acting as contractual party. Number two is where we provide managed settlement-free services. We are beginning to see that model take off.
Are there any standardisation activities in the area of peering federations?
Well within IETF, Speermint, which is one of the working groups dealing with SIP peering has recently provided an update on the underlying interconnection methodology prescribed for scalable interconnection. It defines the role that a federation plays in terms of locate function, signalling function, media function.
Would these IETF SIP interconnection standards pertain to SBC vendors or other parties as well?
It applies to the entire ecosystem of VoIP communications including softswitch vendors, federation/peering providers, SBC vendors, and even the service providers.
Any area where XConnect is active within Speermint?
As a company we are active in registry provisioning aspects within Speermint. At present there is no single common standard being utilized for registry provisioning on a worldwide basis. We think therefore that our contribution will significantly help in uptake of scalable interconnections.
Where are your POPs physically located?
We have three pops. One in the US and two in Europe. Private federations have their infrastructure locally.
There is a trend among peering providers to interconnect their registries. Are you contemplating such an arrangement for yourself?
We have one such arrangement in place already. We are connected to Telcordia registry.
Are we then looking at some sort of consolidation of registry services?
There certainly will not be many registry providers on a worldwide basis. It will be a fairly small number of providers.
What has been the impact of VoIP peering on the telecoms business? You know impact on wholesale minutes business, interconnection costs, impact perhaps on reciprocal compensation, LCR, new services etc etc?
You touched upon all the sensitive points in the introduction to your question. I think while the world is quite comfortable in terms of where we are with regard to VoIP penetration and NGN rollouts, we still are at very early stages. In terms of utilization of new services we are in very early days. So the impact of peering in the second stage, when new services are turned on, remains to be seen.
How does peering help in value added services like video?
There is a crystal clear proposition. Right now service providers connect via PSTN. PSTN is adequate for delivering voice call from A to B. But new services can only be delivered over IP. So either you have hundreds of bilateral IP interconnections for video telephony or you will use federation providers.
Apart from Netherlands and Brazil, do you have federations going in any other countries?
We have them in 3 other countries but we cannot disclose the information.
Any information on that CableLabs RFP in the US?
Can’t comment.
This federation model is not going to scale. Surely you cannot be in 200 different countries. Have you thought of perhaps franchising out your business?
We are quite happy to work with local partners. That is our preferred approach. As an example we don’t provide layer 2/3 services. For that we work with a local partner who is the local IXC or major co-locate where this would make sense.
I had written a post once on possible integration of Facebook type services and ENUM. Do you think the idea is feasible?
With end User ENUM, end user is in control of all the details about how to connect to him: This is my fax connection; This is my SIP address; This is my telephone number; This is my IM address; This is my Facebook identity etc etc. That was the original idea behind ENUM and therefore correctly called User ENUM a.k.a Public ENUM. The last 48 minutes of this call have been discussing carrier ENUM. What your blog is referring to is kind of a hybrid i.e. creating a Public ENUM and linking that with Carrier ENUM. It is an interesting mixed environment. The goal of Carrier ENUM is to enable optimised routing between service providers. Public ENUM deals a lot with who I am, what I am. Data privacy therefore is a big issue in Public ENUM and it is an opt-in thing as opposed to Carrier ENUM which is essentially optimised routing handled by service providers. So they are two very different structures with a lot of regulatory and data privacy issues. I am not aware of any combined activity.
Any new fancy stuff that peering enables that you are working on?
We are working on some of the complex IP interconnection capabilities and the announcements will be forthcoming. Full blown video interconnection is a complex piece. Interconnection of some of the larger IM clouds is also complex.
