::: nBlog :::
Many formal and not so formal organizations, such as the IEEE, IETF and TM Forum have recently come up with proposals for creating standards for the Internet of Things. In addition, companies like IBM, GE and Oracle have invested serious marketing money to their own visions which also include business model designs.
In the telecom world we can see two very different architectural strategies, which worked relatively well for fixed networks and the Internet.
1. Legacy telecoms model: Specify all protocols and future services and implement them in a deterministic way. Enforce rules with rigorous accreditation and prohibit any deviation by customers. This is how original PTT/SONET/PDH/SDH networks were built, and the model was quickly adapted to GSM, which proved to be very successful in a global scale.
2. Internet model: Create an incomplete, skeleton architecture and let future protocols and services evolve among the promulgators and users. Let multiple standards coexists in parallel, letting the more adaptive one win according to Darwinian rules.
The telecoms model’s ideology can be described as ‘father knows best’: user services and devices were designed so that users had very little choice – everything was forbidden unless explicitly allowed by the great phone company. Innovations were mostly limited to making more money with better billing systems, or users circumventing charging pulses with unauthorized terminals. The massive but unrecognized downside, however, was that there was relatively little innovation and development.. in the last 100 years.
The Internet model is inherently open – everything is allowed until explicitly forbidden, blocked or retired as harmful or commercially not viable. This has led to some incompatibility and temporary performance issues, but more widely to an image that Internet never serves you better than at best-effort level. However, this has radically changed the business models of many (non-IT) industries during mere 10 years, and continues to do so.
The Internet of Things, Industrial Internet and Spimes – the digitalisation of all things physical – present us with challenges that are not solvable with existing architectural models. When your pacemaker, car or high-voltage transformer is always connected and adapts to its environment with most of its computing and data storage in the cloud, you’ll want highly reliable operations on a quickly evolving, always-on, global and scalable digital platform.
Most importantly, the new architecture and principles must be designed to support products and services which are not invented yet. When data and algorithms are easily combinable and nestable, these will emerge not only from human innovation but also increasingly from machine learning systems analyzing data flows and patterns.