The path to “lean” operation is one that delivers both better customer outcomes and lower costs. The general principles are no secret, yet they have not been translated to the specific nature of telecoms networks, and hence are not yet adopted in our industry.
As part of the preparation for the Telco and Cloud Quality Revolution workshopin London on 15th March I have been putting together some notes. Here I share some of the principles on which we believe transformational progress to be possible.
We still have a few spaces left on the workshop – contact me if interested and I will send you details.
#1 Humans (and human processes) first, technology later
Whilst technology is important for relieving system constraints, the first step is to understand who has what stake in the quality system, what that system for quality is, and how it is improved (gradually or in jumps).
#2 Take responsibility for learning from failure
Failure is ordinary; we welcome it should welcome it as a natural part of the exploration of new possibility. However, not learning from failure is not good, since failure typically is pointing to systems issues that need to be located. You need to turn the stress of change into embedded learning in the overall management system.
#3 Capture the right intention
You can’t make everyone happy all the time in a finite resource cost, so something has to give. What is the disappointment you are willing to tolerate, and how do you want quality to degrade in overload? Defining “acceptable failure” is the new “success”.
#4 Get everything aligned
There are three distinct forms of alignment needed to get quality right: semantics (your original intention, its written specification, and the operational delivery of that); the service lifecycle (concept to market, lead to cash, trouble to resolution); and the systems (people, process and technology).
#5 Balance supply and demand
Any quality system will be fit-for-purpose when supply and demand are sufficiently balanced at all timescales. “Balance” is not perfect equality, but an acceptable (mis)match. This needs to be done at all the internal boundaries and external management interfaces of the system, not just end-to-end. Hence you need a means of (de)composing quality.
#6 Think flows, not stocks
The essence of “lean” is to move away from inefficient and ineffective “batch” models of system management (and the incumbent bandwidth paradigm is a batch model). For networks, that means using properties of (instantaneous) flow, not averages like “megabits per second”; in turn, that requires you to work with quantities of quality [flow], not quantities with quality [batch].
#7 Measure the right things
“You are what you measure”, since measurement is a prerequisite to management. You need to use network metrics that are anchored into the demand-side and value creation; that also allow for spatial observation (via direct observation or inferred via tomography); and temporal decomposition (to separate dynamic effects from static or structural ones).
#8 Visualise flow and limit work in progress (WIP)
To improve flow you need to do two things: to see it, in the term of the value being delivered, and to minimise the “in flight” resources committed to deliver it. This means we need to represent QoE in user-centric terms of “performance hazards”; have a “network map” to identify common failure modes; and switch from work-conserving queues (i.e. send as many packets as fast as possible and maximise WIP) to nonwork-conserving(*) strategies (that minimise WIP).
#9 Think locally but optimise globally for longevity
Success is a systems property, and our intent needs to be formed at the system level, not the component mechanisms. Systems that are optimised for longevity require an “antifragile” property to overcome catastrophic tail risk of a volatile world. This implies a symbiotic relationship between a “polyservice” network (that provides optionality) and adaptive learning applications.
#10 Focus (only) on systems constraints (process, policy, metrics, tools, etc.) to improve
The first and essential step is to create a programme to “upgrade your quality improvement process”. This means focusing only on true constraints to growth and value. If I raise or eliminate this constraint, how will it let me increase long-term profitability? What do I need to do to change this constraint? Answer these questions, and you will be on your way to a lean telco transformation.
(*) The literature has no consensus on the spacing and hyphens for “non work conserving” – someone who is an expert in English grammar please illuminate me!
For the latest fresh thinking on telecommunications, please sign up for the free Geddes newsletter.