Microsoft has its fingers deep inside the greasy guts of its own Azure cloud computing mechanics right now. The firm doesn’t exactly publicly acknowledge that it lags behind Amazon Web Services in terms of market share, but it is doing as much as it can to make up ground -- it has now invested more than US$15bn in building a global cloud infrastructure and cloud services proposition.
There are currently 26 Azure regions around the planet -- and a ‘region’ in this case can have more than one datacenter in it, so for example in the Azure world we find that American is two regions: West and East.
It is important to remember that AWS is actually bigger than (at least four, some say closer to ten) of its largest cloud competitors combined. The catch up players also include Google, CenturyLink, VMware, Virtustream, Rackspace and IBM’s SoftLayer.
Trillions of transactions in Azure
But market rankings and Magic Quadrants notwithstanding, Microsoft’s Azure engine is running and the numbers are enormous in and of themselves. Executive VP of cloud and enterprise Scott Guthrie recently detailed the size of Azure’s locomotive power at his firm’s Future Decoded event held in London.
- 1.5 trillion messages are now processed by Azure IoT per month
- 800 trillion storage transactions per day on Azure storage system
- > 40% of Azure revenue comes from startups
CEO Satya Nadella spoke at the same event as Guthrie to explain how he regards Windows 10 as a ‘first step’ towards Microsoft providing an Operating System-as-a-Service. This is good keynote fodder, but what it really refers to is a world where all kinds of devices from IoT sensors to desktop computers use some element of Microsoft DNA to drive their intelligence.
With this in mind, Microsoft will is now moving to offer a product it calls Azure Stack (an evolution of what used to be Azure Pack) where customers will be able to get elements of Azure and other Windows-type operating system power including Windows Server and the System Center operations management system.
Where does all this lead us then? Part of the new development of cloud computing inside Microsoft will lean towards what the firm has called ‘the intelligent cloud’. Beneath the branding gloss here, this refers to cloud computing resources with a degree of analytics and machine learning i.e. the kind of technology that will produce computational elements like ‘recommendation engine’ technologies that suggest additional purchases for us when we shop online.
The connected cow
At a fully blown level, machine learning obviously extends far deeper and wider than online shopping to bring us towards so-called ‘deep neural networks’ and Artificial Intelligence.
Microsoft’s Guthrie uses his ‘connected cow’ presentation to explain how a farming installation in the Netherlands has been established with cows wearing pedometers to monitor the movement and motion of the animals throughout the day.
This story is meant to explain that a farm (like any business) needs to maximize output (which in this case are milk and beef) under the constraints of fixed assets. Being able to track the asset (a cow) means that the farm can control its business better. Hence, electronically tracked cows are a better business proposition. In theory at least.
Farmers need to know when cows are on heat and reach the state of what we call ‘estrus’, when the probability of calf production resulting from insemination is highest. With a tagged cow the farmers can plug their data into a cloud based system to a) perform analytics and tell them when to push insemination activity forward and b) share data on cow health and start to track the emergence of diseases and other issues common to the species.
A massive proliferation of cloud tools
What all these leads us to are the massive proliferation of cloud tools and services that allow us to interact with the data in the cloud world. For Microsoft this means products like Microsoft Power BI. This is the kind of technology that will allow users to build live dashboards and reports that work on the data being processed. It can even be connected to live IoT data to see what is happening to Internet of Things as they actually live.
If a company (a farmer or any type of business) wants to see timeline-based changes in data so that reports can be made from the data being processed then now they can perform this calculation. This intelligence can, in turn, be fed back into the systems that are producing the raw core data (such as the cow pedometer for example) and at this point the system can start to get smarter about what it needs to do with the system it is managing – and at this point we can call this machine learning.
Deep neural networks
As cloud-driven machine learning now evolves we have almost forgotten that it was only a decade or so back now that Artificial Intelligence was the stuff of science fiction books and movies. We hear about so-called deep neural networks in computing terms, but we use the term ‘neural’ as a direct reference to the complexity of the brain and its synaptic connections.
Microsoft may not win market share in the cloud wars through sheer infrastructural investment, but it will win kudos, customers and even cows as a result of accessibility to the right kinds of tools and services – and this is precisely where many of its cloud development is now focused.
Didn’t you think that cow pregnancy-tracking pedometers were part of the Internet of Things and its cloud analytics-based machine learning intelligence offering before now? Moo to you then.
