The user experience of packet networks is an instantaneous phenomenon. The common industry performance metrics fail to reflect this fact. As a result, we cannot adequately measure, model or manage the user experience. Are you ready for The Big Decision?
Here is the reality everyone in the telco and cloud industry faces, like it or not:
- Data networking is inter-process communications, i.e. it puts the “distributed” into distributed computation.
- From a distributed computation viewpoint, networks attenuate quality whilst translocating information between computations. It’s what they do, and is the only externally observable effect.
- Quality attenuation frames the network resource as a “privation” (i.e. a thing that has an “ideal”, and can only “impair” from that level, like service availability from 100%, or silence in a noisy café).
- Quality attenuation is the instantaneous network impairment that each packet experiences, which is manifested as discrete loss and/or continuous delay.
- ∆Q is a mathematical construct that unifies loss and delay into a single metric space (the “complex numbers of probability”), and hence quantifies the quality attenuation.
- This single object (when suitably expressed) can be meaningfully composed as the ∆Q algebra (the fancy term is a monoid).
- The quality attenuation (“impairment”) of the packets can also be related to the quality attenuation (“impairment”) of the user experience via the ∆Q morphism.
- This compositional algebra and morphism allow you to reason about the network performance by the horizontal and hierarchical “addition” of resource demand and resource supply, and to compare these.
- Such a calculus of network resources is an essential prerequisite for answering classic engineering questions: feasibility, safety margin, performance, cost and risk – and any trade-offs of these.
- These techniques open up the possibility of performance and cost engineering of complete digital supply chains, and enable new arbitrage and assurance models.
- The same mathematics and science can be used to design and engineer new network mechanisms that provide assured user experiences, even in overload.
- This new “digital logistics” with managed and predictable outcomes is an opportunity to exceed the “possibility zone” in which existing industry players operate; it is “impossible” in their network cosmology to manage quality in saturated networks, so vast market opportunities are invisible and untapped.
- The result is a potential for radical business transformation, comparable to how “lean” and “six sigma” concepts transformed manufacturing and service industries, or how containerisation and logistics revolutionised shipping and distribution.
Either you have the ability to work with the instantaneous network performance, or you don’t. There is an inevitable industry “upgrade” from old metrics for periods to new ones for instants. The inescapable decision you face is as follows:
- Engage with this reality now, whilst the science, tools and technologies of instantaneous network management are embryonic. You do this in the hope and expectation that a “winner takes all” of software-driven resource trading network architectures applies. This is a commercial bet that a new multi-sided model will dominate cloud, and displace the current telecoms circuit model over time.
- Engage with it later, in the hope of being a “fast follower” after others have borne the cost and risk of initial development, and met the inevitable initial setbacks and failures. This is a bet that traditional one-sided telco models will dominate cloud connectivity, even as quality-centric architectures emerge and evolve.
- Ignore it and do not engage, knowing there is a risk of having a “Nokia moment”: you might suddenly go out of business as others wildly outcompete you in the market for distributed computing services. This is a bet that the existing business can be consolidated, milked and optimised long enough for you to collect your bonus and pension (without shareholders noticing and firing you).
As far as strategic choices facing tier 1 network equipment vendors and telcos are concerned, this is about as big as it gets. Get it wrong, and it is like missing the transistor, microwave radio, laser, or fiber optics. The shift to total software control over all network resources (and hence user experience) is fundamental and unavoidable.
To give you some hints as to what the right decision is:
- The increasingly dynamic nature of “faster” networks makes instantaneous contention effects the dominant factor in user experience.
- 5G is at risk of technical shortfall or failure, since it lacks the necessary instantaneous quality management.
- SD-WAN is already arbitraging the current telco model from the edge, by managing instants better than the operators can.
- When the new “high frequency trading” algorithms move into the network, it is game over for existing players.
If you have anxiety about making the wrong choice, and would like help with this very big decision, please get in touch.
For the latest fresh thinking on telecommunications, please sign up for the free Geddes newsletter.
Read It Now
