The more VoIP changes the telecom business, the more predictable it gets. During a recent conversation with the heads of sales and operations for VoIP peering service provider, Infiniroute, I was reminded of the evolution of the classic ISP peering model. Just as technical necessity and economics drove the ISPs to directly connect through peering points the same holds true for the carriers now using VoIP. Access is the key to the marketplace. Interestingly though, many of the international incumbent carrier networks have been architected to not be openly accessible, but rather have one main gateway typically using H.323 that acts as the interface from their network (and user base) to the rest of the world. Lets see if IP, ENUM, and SIP allow that architecture to survive.
The issue for the incumbents is that they need to move to IP, but dont want to change their revenue model just yet. They can somewhat accomplish this by using H.323 and establishing only direct peering relationships with other carriers on a bi-lateral basis. This design is a product of timing, limitations, and a touch of sunk costs.
Many of the international carriers established these VoIP gateways several years ago when VoIP was brand new as a means to leverage the benefits of IP for the purpose of provisioning voice circuits to other carriers. Back then H.323 was the dominant protocol. Thats the timing part they were early adopters in a sense. The limitation part of H.323 and subsequently the reason why practically everyone in the world is moving to SIP is SIPs ease of interoperation. SIP is a very lightweight protocol meaning there is not a lot of overhead, bulky code, or processing involved. H.323 has more of a traditional telecom background whereas SIP comes from the IP world. Most importantly it enables dynamic provisioning through SIP addressing. That makes SIP more efficient, easy to use, and ultimately cost effective.
This is not lost on the incumbent carriers either. What is very telling is that most of their internal architecture is based on SIP because it makes their own network easier for them to manage. So why the H.323? The lack of dynamic provisioning on the carrier/customer level for buy/sell relationships makes it difficult to connect, route, and reroute to them and requires in many cases personal negotiation for call termination on their networks. This keeps them in control and their wholesale business from being further commoditized. For most non-incumbent wholesale carriers SIP is dominant because they need to make it easy for others to connect to them since the middle-man business is so cut throat and margin-less these days. SIP is the next step in the path to ENUM and once that point is reached devices can communicate directly with each other. Eliminating long-distance puts a dent in the incumbent revenue model.
The sunk cost in the H.323 equipment for the incumbents is somewhat of a factor for things remaining the way they are as the incumbents dont want to remove it if it is generating revenue and works properly, but the provisioning limitation provides a convenient technical reason why open-access cant really happen and subsequently protects the revenue.
Keeping the past alive in this way presents an opportunity for Infiniroute, which provides route optimization over public and private networks as well as signaling information for direct routing between H.323 carrier gateways and H.323 to SIP conversion. Their model works because the incumbents own the end user access and therefore the traffic to and from those users. Getting all of these VoIP islands to directly connect and avoid any unnecessary hops enhances quality and lowers costs. Infiniroutes direct routing facilitates the layer 7 aspect of classic ISP peering by identifying the gateways and then it looks for the optimal media path for transport.
Once the signaling request is made our system determines the best way to route the call based on quality, available capacity and cost, states Neal Axelrad, VP Sales for Infiniroute. This may be over the public Internet, which is our strong point, but many of the incumbents already have direct connections between each other and they are typically Layer 2 Ethernet. Where quality is concerned a private network connection is usually always preferred, but well optimally route them either way.
That says a lot about how ALL network operators think.
These private peering connections between the big incumbents are very similar to the Seven Sisters of ISP folklore. They privately connect, cut special deals, and dont open their networks. Although this pattern is repeating now in VoIP the fact is that it might not be that easy to keep change from happening, or even slow the pace. All of the technology the VoIP carriers are using is readily available for any medium to large enterprise. That coupled with pure-play transport (mainly Ethernet) carriers with no Layer 3 service model and all of the pieces are in place. This means the enterprise user base can step around the carrier, build their own VoIP network, peer with other enterprises and the minutes (and revenue) go away. Its not that difficult, its already happening and as it begins to really take hold the addressable market for Infiniroute and other VoIP Peering service providers will expand by several orders of magnitude. Anyway, lets face it, back in the good old days of IP transit (1995) a who lesale meg of IP was $1,000. Today in 2005 its $10. Oh well, so much for protecting the top line revenue with a closed network model.
In the grand scheme of evolution, whether its VoIP, or a living species, eliminating inefficiencies is natures way. We are all tied to a logical genetic structure in what we create as it is the way we were created. As long as we are able to put aside conflicting political and legacy economic pressures to resist the evolution the end result will benefit society. Just because it is good for humanity though doesnt mean its good for the incumbents bank accounts, but in the end no one can stop the VoIP volcano from permanently changing the landscape.
Mismatched protocols and creating VoIP islands is not the optimal path. What this global voice organism wants is seamless interconnection with all of its parts, but will the enterprise VoIP wide-area networks eventually be interconnected to each other and eliminate that part of the business of the IXCs in the process? Since many enterprises are basically starting with a clean slate and have no interest in generating revenue from voice as they see it as an expense, there is a very good chance that that will happen. Its all a matter of time. IT
Hunter Newby is chief strategy officer at telx. For more information, please visit www.telx.com.
If you are interested in purchasing reprints of this article (in either print or PDF format), please visit Reprint Management Services online at www.reprintbuyer.comor contact a representative via e-mail at [email protected] or by phone at 800-290-5460.