GOC

Once every now and then a technology tsunami hits our shores. Like real ones, the warning signs are subtle, but the impact profound. Virtualization technology was one such event and the change on our computing and communications landscape is still unfolding.

One of the historical complexities of operating managed services was the development of large-scale, multi-enterprise platforms to deliver services. One could argue the only difference between an enterprise PBX (News - Alert) and hosted communications service is simply the ability to service many different organizations from the service provider switch versus a single enterprise from the PBX.  Server and application virtualization takes this well-worn paradigm and throws it away. Using today’s virtual environments, it is now possible to deliver any enterprise application as a managed service, and the very makeup of service provider infrastructure is being redesigned before our very eyes.

To get comfortable with this new framework, consider the concept of the hosted IP PBX. It was less than one decade ago that companies like Broadsoft, Sylantro, NetCentrex and others developed platforms to deliver ‘hosted PBX’ services. The basic idea behind all of these implementations was a multi-organization platform that allowed service providers to deliver PBX functionality in the cloud as a managed service. These companies survived, but stiff competition from the premises-based PBX companies with well-established channels kept adoption modest.

Even while these hosted services were developed, PBX companies underwent an evolution of their own. Building on readily available, standard SIP software stacks and hardware endpoints, many companies like Asterisk (News - Alert), Cisco, Sphere and Microsoft started developing PBXs as applications on standard operating systems. Today, for example, Asterisk runs on Linux and Cisco Call Manager runs on Windows Server. 

And this is where the story gets interesting. What happens if you take a robust enterprise PBX like Asterisk or Cisco (News - Alert) Call Manager and create a unique copy per customer using a server virtualization platform in the cloud? It looks a lot like a hosted PBX service, except it was wholly delivered without a special purpose, multi-enterprise server provider solution of old. If you think I’m crazy, just search the Web for virtual PBX and see how many hits you get. This model has been in production deployment for close to a year at some small companies, and I have even heard rumors of offerings along these lines from some of the larger PBX companies.

I could be wrong, but I can see this model really gaining ground and going well beyond the PBX. Why spend precious software development and architecture cycles on monolithic, high-availability systems when the virtualization platforms are doing that heavy lifting? Why not just build really great enterprise apps and host virtual instances of them in the cloud on large utility computing environments? As someone who operates service provider systems, the idea of spreading my risk across many instances of an application is more appealing than a crash at a shared application layer that takes down thousands of customers.

Yes, I know this means you have to place your trust in the virtualization technology, but based on the billions being invested in that area, I think it’s a rational wager. And when I consider the flexibility of integrating additional services or installing new code for a single customer instance versus a shared platform, again, I see the positives.

 There is no doubt, virtualization is not a panacea. There are many challenges associated with managing potentially tens or hundreds of thousands of virtual environments, and the technology for this purpose is still in its infancy. But I believe solving those kinds of challenges is more virtuous than building another high-availability service provider platform the old way. The benefit is broader and leads to a future where any application can be hosted as a managed service.

When the water finally recedes, it’s hard to say what will be left of the traditional service provider infrastructure. But of this I am sure – those engineers at “Ma Bell” will not recognize it.