I was recently asked how telcos might come up with new business models for a world where all resources are under software control. The core idea is to match network supply and demand in space and time (and at all timescales). I’ve typed up my notes for the curious to critique.

The traditional supply-centric model

Today’s telco is more like a static utility ‘pipe’, the way that gas, electric or oil are delivered. It sells direct access to raw network mechanisms, and users buy circuits that offer a fixed bandwidth in space and time.

The application experience risk not generally a concern of the network operator. Buyers of circuits get a fixed or low variability quality with bearers based on TDM, ISDN, ATM, MPLS and Ethernet technologies. This pushes the QoE risk management onto the buyer, who has full control over timing of data put into the circuit, and how much goes in.

Where services are delivered over broadband using Internet Protocol, then there is more variable quality. The control over the experience is internalised to network, but with limited user ability to direct the resource allocation. As a result, it is typical to use over-delivery of quality to manage QoE.

The billing in today’s model is very simple: it is for a quantity of resource used over time (e.g. a Call Detail Record for a phone call).

The future demand-centric model

The world we are heading into is less like a static ‘pipe’, and more like a dynamic financial services ‘resource trading platform’.

Rather than selling raw mechanisms, telcos can sell (proxies for) different levels of QoE outcomes and associated business risk. They might do this by selling “application erlangs” (e.g. “Lync seat”) or “software-defined circuits” (e.g. ISDN line replacement). This abstracts the end user value from the network implementation, and allows for more dynamic implementation flexibility.

This demand-centric service model allows for managed user QoE risk, with tiered QoE levels linked to ability to pay. The billing model evolves from pure quantity to variable “quantities of quality” with different resilience levels. The underlying delivery mechanisms are no longer visible, being abstracted away.

A new kind of model will need new kinds of operational and back office systems. For instance, we may meter these services by “Quality Data Records”, which requires billing to be come quality-aware.

The implantation of these new kinds of service also need network capabilities. We need both underlying high-frequency (“SD-QoS”) and low-frequency (e.g. SD-WAN/SDN) resource trading platforms for “ballistic” and “elastic” timescales. These in turn need new metrics, models and mechanisms to be able to relate resource allocation decisions to telco-managed end user QoE risk.

New commercial models for a new paradigm

What is perhaps most interesting is how these new technical delivery methods enable new commercial models.

The simplest example is the ability for operators to “steal” ideas from the physical logistics world. They can become supply chain performance managers over their own and third party connectivity. They can contract and manage the resources to deliver specific QoE and cost outcomes. This requires them to measure quality at interconnects and management boundaries using high-fidelity metrics that are strong QoE proxies.

Another kind of business model is performance arbitrage, of both the operator’s own and third party assets. You can think of this is “next generation least cost routing”, but in time domain as well as space. Every application has a quality need, and the task is to exploit the “gap” between the (excessively high) underlying network quality and the (lower) application requirement.

You can only do this arbitrage if you know what the the application need is, i.e. a demand-centric model. Its execution has both “exploit” and “defend” sides. How can you exploit the over-delivery/under-pricing of quality of your rivals? Where you are offering mispriced quality? Can you join together multiple cheaper assets to create a substitute service?

As you get more sophisticated, you can start to offer services to manage “portfolio risk” of application performance, just like a financial investor cares about portfolios and not individual stocks and bonds. These more advanced services balance QoE risk and cost management for large enterprise clients.

The world we live in is an imperfect one, and things can and do go wrong. Once you start to understand your resources and their costs well, you can offer performance insurance services for any residual risk of application failure. This requires you to use your trading platform’s risk pricing “network insider” information advantage.

Finally, as you get to the highest levels of sophistication you can offer resource futures and options trading services. These create early price signals for future demand, so as to better direct physical data transmission supply investment. It also helps to flush out which public clouds and data centres are going to be the “gravity wells” of future connectivity demand.