IntelePeer: Reverse Engineering Telco 2.0
IntelePeer is a Californian startup that should worry telcos quite a lot. It's constitutive of Telco 2.0 that we want to identify the key telco capabilities and assets that are hard for competing over-the-top players to replicate, make them available in a useful form, and then monetise. But one of the risks here is that other players - OTT'ers, software developers, big clouds, big IT services firms, device makers - start to replicate these faster than telcos start to develop them.
What is IntelePeer and what's important about it?
They describe themselves as a "communications as a service" company, which is another way of saying that they are actually quite telco-like. After all, telephony was the original software-as-a-service product - all the intelligence is at Telco HQ, and you just dial in, and you mostly pay for what you use. IntelePeer's products include essentially all the elements of a voice network - they have extensive peering relationships for VoIP and also TDM transport, they have a hosted switching infrastructure with an API for Web-voice integration (the AppWorx platform - note that they call it a platform!), and they have a so-called SuperRegistry, which is essentially a private ENUM deployment.
This gives them the ability to serve even complicated platform/VAS and Voice 2.0 applications; the registry is especially interesting, because one of the main barriers to better voice and to the wider re-use of telco data assets for identification, authentication, and sociability is the difficulty of mapping telephone numbers to other services and vice versa. The official answer is ENUM, a project, led by the IETF, which aim to replace the existing telco system with one based on the Domain Name System.
Essentially, you'd be able to query the DNS for a telephone number and be given the IP address of the end-point with that number, or instead be referred to another domain name, for example a SIP ID; in the other direction, it would be possible to register a meaningful name for a service or resource identified by a traditional e164 phone number. And you could have different kinds of records attached to a name or number - so others would be able to know where to look for other things you might be using beyond steam voice or IP. To cap the lot, it would also provide for federation (just because your phone number is assigned by Vodafone, this doesn't mean that your ENUM record couldn't point your answerphone at Me2Me) and for user ownership of their identity.
Now, we're going to have to do this anyway with the move towards all-IP networks; IMS depends on ENUM, but flat-architecture IETF SIP networks will need it as well if they want to use telephone numbers at all, and even XMPP uses the DNS to make its interconnection work. But it's especially critical for applications that use customer data, or that need to interact with other services. However, the progress has been very slow indeed, and it's no surprise that we're beginning to see efforts to build private deployments.
So there's an effort to replicate telcos' "Who are you?" and "How are you?" capabilities. Of course, the real source of differentiation for the telcos isn't the technology or the infrastructure, but the accumulated data pile - but as soon as people start using a rival system, this advantage will start to diminish.
Then there's the network infrastructure; BT's Matt Bross said at the last Telco 2.0 event that he wanted the company to use the infrastructure built for global voice interconnection to develop new applications. Telcos have always been big on interconnection, even if some resisted peering at IXen for years; equally, they enjoy claiming that only they can provide "carrier grade" performance. (We remember watching an IMS video-sharing demonstration one 3GSM that was almost totally nonfunctional due to very high latency, while the demonstrator explained that only IMS-enabled telco networks could guarantee a good user experience.) IntelePeer is buying dark fibre and building an IP MPLS backbone network, whilst also aggressively seeking new peering relationships - they aim to eventually become transit-free. Creating their own MPLS core and peering relationships gives them more telco-like control of costs and quality - part of the answer to "how can we reach you?"
And they're also working hard on making these things available as APIs so their customers can spawn their own developer communities.
According to IntelePeer, about 95% of applications that use their services are interacting with them through their REST Web service (the rest are using a PHP library); these are mostly business-intelligence, CRM, and related applications. They recently signed a partnership with Microsoft to integrate their platform with MS Live applications - and more broadly, to get access to MS's huge base of both customers and developers.
When we spoke to IntelePeer management, they were keen to depict telcos as partners and software developers like Ribbit and IT service companies like IBM or BT Global Services as competitors. But the main resource they need from telcos is raw bandwidth. Being a trusted partner is all very well, but if you are their "bit pipe partner" it may sound a little hollow! At the moment, they may be a big customer; but structurally, their strengths directly challenge many of the key assets telcos bring with them to Telco 2.0.
And, as usual, a significant chunk of their infrastructure uses the Asterisk open-source telephony platform; as we keep saying, every Voice 2.0 player we see runs Asterisk, even if IntelePeer also uses a lot of big-iron Sonus Networks IP voice switches.
[Ed - we'll be debating the issues raised in this article at the next Telco 2.0 event on 6-7 May in Nice.]