|
 The vision of convergence -- anytime, anywhere communications -- has been the holy grail of the telecom industry for almost 20 years in one form or another. Converged services combine voice, data and video and deliver them across a span of public, private, wired and wireless networks in a way that is seamless and consistent to the end user. The problem with holy grails is not so much that they're so hard to find, but what we might miss out on while we're looking for them.
I remember the feeling I got back in the early 90's when I first heard the crazy idea of putting voice communications on the Internet. My boss at the time got this odd, pinched look on his face and said, "it would never be as good as today's voice systems, who would want it?". But I'd caught a bug, and I spent the next 7 years working with VoIP, and everything I've done since has been connected with this vision of convergence.
And of course I wasn't the only one. The excitement didn't come so much from the idea itself, but from the possibilities it suggested -- if it was possible to put voice over IP, what else might be done? In a lot of ways that excitement has been driving research, development, investment, and imagination in the telecom and networking industries for almost two decades.
From a technology perspective, the key to delivering converged services is the ability to separate the application layer from the network layer, essentially making it possible for the application server to deliver a consistent experience to the end user regardless of the user's location or the networks in between. The challenges in accomplishing this are predominantly network ones: does the application have sufficient bandwidth, can it be monitored, can it be billed, is it secure, is it sufficiently free from packet loss and delay, can it adapt to changing conditions in the network without affecting the end user experience, etc.
This can be complex enough with a single media, single network service, but it is exponentially more difficult when multimedia services (for example, services which include the interaction of voice and data) and multiple networks are involved. Since the mid-90's there has been an incalculable amount of time, effort and money devoted to overcoming the obstacles and making the unversal network a reality. But in many ways, the industry is no closer to a solution.
One of the more spectacular recent efforts has been IMS (IP Multimedia Subsystem), a sprawling initiative originally intended to bring IP services to mobile users. In a recent technology brief , Tom Nolle described IMS as "the closest thing to transcendentalism that exists in networking, [which] captivates millions and has deployed essentially nothing."
What caught my interest in Nolle's paper was how he skittishly raises the possibility that the idea of true convergence in service delivery might be fundamentally contradictory under the current network models involved. He narrows the problem down to one of control of network resources -- specifically that there is no common control mechanism between networks which traditionally operate on a free-peering, best-effort delivery system and those that depend on deterministic provisioning (such as enterprise networks or settlement-based networks). Neither of these models can change, and the feasibility of an intermediary product to hybridize them is problematic at best.
So Nolle complains about the current impossibility to create services that require "real pan-provider, multi-technology and vendor, reserved resources", and he's quite right. But what if there is no solution without completely reinventing network infrastructure, which is pretty much throwing the baby out with the bath water? Does it ultimately mean that convergence is a dead end?
Somehow I don't think so. Convergent services are a reality today, so much so that we take them for granted: voice over IP, mobile browsing, VPNs, and so on. Others are emerging, such as fixed mobile convergence (FMC), software as a service (SaaS), presence based routing, etc. In fact it's hard to think of any area in the technology sector today that isn't convergence based.
The difference is that these are examples of "limited" convergence. To varying extents they are niche applications: they don't fit into a larger architecture that will allow them to interact and use common resources in the ways we envision. However, unlike the applications which do fit into that larger scheme, they've actually been deployed and can be profitable.
Contrast this with the attempts to create panacea solutions that become quickly mired in their own complexity. To continue with IMS as an example, as a pan-network architecture it seems to be collapsing under its own weight. Vendors and service providers have been able to cherry pick some aspects of IMS and deploy them, but it's a safe bet that no one will ever commercially deploy a complete IMS architecture (sorry).
Einstein famously said, "Make things as simple as possible, but no simpler," and one could argue that only these niche applications have the potential to be viable. The complexity of "any-service, any-network" is economically out of reach, like trying to throw a baseball to Saturn. With niche applications, we're just trying to get it across the plate.
Which brings me back to the holy grail idea. If a universal solution isn't achievable and niche solutions will increase complexity and put the vision of convergence even further out of reach, where do we go from here? I think there isn't an answer needed so much as an attitude. They say it's the journey that's important, not the destination. We keep searching for that grail, but at the same time we accept that we might never find it. This isn't discouraging, because if we stopped looking we'd miss out on everything we encounter along the way.
 |
Show/hide comments
Show/hide comment form
