In my September column, I focused attention on the diversity of new capabilities spawned by the power of cloud computing – Machine Learning, Internet of Things (IoT), Robots, Big Data/Big Analytics, Artificial Intelligence (AI). The contemporary business board now needs to focus on this diversity of fresh capabilities that are transforming how business can be done. They are becoming so powerful that every board director should be literate in them.
In September I visited the excellent Robots exhibition at the Science Museum. There I met Kaspar the creation of a team at the University of Hertfordshire. Kaspar is a child-sized robot able to socialise constructively with autistic youngsters. These are early days for AI, and for robots. But Kaspar emphasises my point. There are powerful new capabilities now in play: their potential business contribution is developing fast.
Earlier, travelling along the Regent’s Canal, I recorded a podcast with Horizon. In it I draw analogies between the impact of the canal revolution in the late 1700’s/early 1800’s and the AWS-lead cloud computing revolution some two hundred years later in the very late 1900’s/early 2000’s. And I point out that the canal-focused entrepreneurs of the early 1800’s failed to notice a new competitor – the railway.
Two hundred years on, is there a challenge to the business models of the cloud computing majors as we know them? Data storage, data processing, data networking are now all software definable. There is consequently an emergent reality called Big Software. A focused coterie of innovative tech players are exploiting this new reality to provide the means to quite radically transform our practice of Dev/Ops and Infrastructural Operations – the Trevithicks and Stephensons of our era! (rail pioneers for non-train spotters – Ed)
Firstly the emergence of bare metal services, as they have been dubbed – raw computing power at the server level available to be sourced on demand, on a spot basis. One example is the US start up Packet investing heavily in facilities that bring compute power close to developers, exploiting existing networks, and by avoiding networking overlay or multi-tenancy, allowing developers to quickly deploy an instance using standard DevOps tools – and integrate software-defined wide area network (SD-WAN) capabilities enabling integral exploitation of the networks. Lower costs and reduced latency.
The fundamental revolution in play above the bare metal layer is the development of focused, unitised software architectures that allow the potential complexity of the Big Software ’mass’ to be managed as an assembly of units that are operationally robust and, importantly, self-referential (so that assembly is in essence an automation exercise). Data storage, data processing, data networking managed in a holistic manner – exploiting the underlying ‘bare metal’ servers with maximum efficiency, maximum effectiveness.
Thus new operational concepts, including: Containers: a format that encapsulates a set of software and its dependencies, the minimal set of runtime resources the software needs to do its function. Docker: providing an additional layer of abstraction and automation of operating-system-level virtualization on Windows and Linux. Kubernetes: a platform for automating the deployment, scaling, and operations of these containers across clusters of hosts.
The essence of this new world is that it allows a focus on distinct units of software (maximum re-use wherever possible), that can be easily assembled (fast move into the Ops end of Dev/Ops), more easily altered and adjusted (the Dev end of Dev/Ops). This in a commercial world where the emphasis is on market responsiveness and agility with tight control of the cost of operations thus enabled.
A lead pioneer is the US venture Canonical, working primarily with OpenStack and its open source software operating system Ubuntu. Juju is its open source application modelling tool – ‘its mission is to provide a modelling language for users that abstracts the specifics of operating complex big software topologies’. Through its development of Juju (Cloud deployment, integration, scaling, upgrading), LXD (Pure-container hypervisor), Ubuntu-Openstack (optimized and interoperability tested openstack packages) – and Ubuntu Server (hyperscale server operating system) Canonical provides the complete operational stack to exploit stripped-down ‘bare metal’ compute capability. Exploiting Metal as a Service – MaaS – to deliver agility, market responsiveness, competitive edge.
The railways, as they gathered momentum through the 1800’s, did not destroy the canals – as articulated in the podcast there was an extensive period of (tough) competitive coexistence, the canals sharply changing their business models, losing profitability. So will be the case now, with these innovative approaches to managing the challenges and complexities of Big Software carving out their market spaces while the classic cloud compute players adjust re-position and loose margin.
Two hundred years on from the competitive battle between the canals and the railways as the nineteenth century speed on, we will now see classic cloud computing business models challenged in a major fashion by these new capabilities spawned from Big Software as the twenty first century speeds forward!