The Disconnect Between Services and Networks

We think of OTT (Over the Top) as competitors to the carriers these days, but before the Internet went commercial the network was designed to be over the top of the carrier network. If you think of customers in the abstract, which in theory has implications for peer-to-peer as well, you can see not all services are over the top.

So here is the issue. The value of a network is in its endpoints (even if I think the network’s cost-value is based on its efficiency). Many OTT services give the carriers credit for just being there and simply expand their processing and transmission without any regard to the impact on the network. That’s even though the people in charge of the network are experts at aggregation, which I bring up because it highlights a flaw in the discussions about Network Functions Virtualization (NFV) and the nature of the operators.

First, let me sing the praises of the Operations Operators. You may feel – rightfully – that your needs are not getting met, but like a coal man on a locomotive, the guys running the network very rarely have a network failure.

The Sum of the Parts may be getting the shaft, but the greater whole runs well, and right there is the disconnect: Is your need for service being reflected in the requirements on the network to manage resources? The proponents of NFV would say that is exactly what they are trying to do –align services with network efficiency – but without a solid network API the abstract service does not equate to the manageable network requirements. It’s like the child of a tax accountant, asking their parent to play on April 14^th … the desire maybe there but the attention won’t be. If NFV does not have a resource allocation method that binds a service to the network, then the idea that services need network awareness is moot.

So let’s see if we can name a service that the network operator can offer, that cannot be replicated (probably easier from outside services).

Location? GPS and Google have shown that location can be determined in many ways, and they have added information to the point where network location information looks pretty sparse.

Billing? First off, billing is an adjunct and not a network service, so even the carriers’ billing information is in effect OTT. Second, there is nothing that attaches to the network billing features. The decision to serve or not serve is Boolean and comes in the abstract. Applications have expanded the billing models with “freemium” and other subtle nuances that require a dial rather than an on/off switch.

Resources (aka Bandwidth)? In the end, this is the only network resource that matches well to network Ops in delivering something new. The reality is that even this capability competes with alternative solutions that use compression and redirection (such as VPN and Routing Protocols and, of course, alternative networks).

In the final analysis, I can only think of one way NFV can bundle network and services, that being in the service of Disaster Recovery, delivering on the fear of a network being swamped.  And now the disconnect is on the other side of the line, as Ops has done their work with consideration of failover and maintaining the whole network.  Delivering an over and above service is not impossible for them (Fiber and SONET are examples); the issue is delivering it as an On Demand service.