Telecoms is reorienting itself from selling bandwidth (i.e. quantity) to managed latency (i.e. quality), as a new Vodafone website article demonstrates.
I do like it when Christmas comes early! Vodafone has put out a rather significant piece of PR about ∆Q. For newcomers, ∆Q-based network technology is to 5G what CDMA was to 3G: core innovation based on algebra that hardly anyone understands, but absolutely everyone needs.
The article focuses on the first step to scientific network management, namely measuring with ∆Q metrics (the prerequisite for later modelling and management). Here are the article highlights for the hard-of-clicking:
- “Vodafone has built Europe’s largest Next Generation Network, passing 99 million households…”
- “Often when a customer complains of poor broadband it’s actually a latency problem.”
- “So while we’re working on upgrading the physical network, laying the foundations that allow us to deliver next-generation speeds, we’re not resting on our laurels. Our next focus is to deliver next-generation quality.” (my emphasis)
- “Although ∆Q is still a relatively new analysis technique, this project has already given us much greater insight into the different aspects of network deficiencies and events we are able to detect.”
- “The next phase is to examine ways to deploy the ∆Q probes in a more ‘frictionless’ manner…”
- “It’s early days, and DQ is just one of a number of things we have in the pipeline, but we’re excited about the difference it will make…”
As context I would like to offer you the following email extract (with readable text below). It was sent to a public list by Dr David Reed of MIT, one of the coauthors of the (in)famous “end-to-end” paper on Internet architecture.
What it says is: “I suggest we stop talking about throughput, which has been the mistaken idea about networking for 30-40 years. Almost all networking ends up being about end-to-end response time in a multiplexed system. Or put another way: ‘It’s the Latency, Stupid’.“
Even those whose names are most notoriously attached to the “fat pipes and overprovisioning” mantra are finally coming round to the reality that the value of networks is in constrained latency, and not unconstrained bandwidth. I can already hear angels drunkenly toasting the repenting sinner with Yuletide mulled wine.
So why is the global Head of Fixed Access at a £47bn turnover telecoms group so excited about an algebra of latency? Well, put yourself into his shoes. Your multinational company operates in 26 countries, and inevitably each network is assembled from both your own assets as well as those from third parties. The customer will also often have a LAN on their property, too.
You have to engineer a user experience for multiple access and application products, and do it over many different bearer technologies, and with many different supplier partners. There’s a lot of variability, and which drives complexity and cost. You are also naturally keen to avoid any “blame game” when things go wrong, as life is prone to do.
On top of this, you’re being asked to support the delivery of low-latency “slices” for 5G, distributed cloud and “edge compute” services, and the much-hyped SDN/NFV technologies. This all has you ordering extra worry beads from Amazon, since there are rather of lot of ways for these ambitious endeavours to go awry.
The way we simplify all these problems, and have a safe night’s sleep, is with scientific management based on reliable engineering and robust science. That’s where ∆Q-based quality metrics come in. You can think of ∆Q as a unified probability model of latency and loss. (Its posh name is “quality attenuation”, but it could equally have been called “latencyloss” as one word!)
What’s super special about ∆Q is that it lets you “split up” the “latencyloss” into three independent components: Geographic delay, Size delay, and Variable delay. When you express the network quality this way, a piece of mathematical magic happens: you can “add up” the “latencyloss” along the network supply chain.
Now do you get why this is so exciting to someone who has to build the technical quality SLAs at all those boundaries? They have to “add up” to a working service!
If you issued a press release that you’d figured out how to add up the power consumption of your network equipment, you’d probably be looking for a new job pretty quickly outside of telecoms. And if you proudly boasted your clever scientists could figure out the total weight of the equipment on the cell tower, you’d be laughed at.
But the immature state of packet networking is such that “adding up” demand and supply of performance is a genuine breakthrough. Vodafone has clearly taken the industry lead, and is strongly hinting the challenge is to industrialise these new techniques with “frictionless” deployment.
Those of us “in the know” are aware that several other major industry players are taking a very active interest in ∆Q methods and technologies. When your competition is doing next-generation quality, you wouldn’t want to ask Santa to deliver you last-generation quantity, now would you?
For the latest fresh thinking on telecommunications, please sign up for the free Geddes newsletter.
Read It Now
