Cloud computing powers the data economy – of course, but how?

The value of data is unlocked when organizations are able to collect and “refine” mountains of it. Refined data is the fuel for running Machine Learning (ML) algorithms to make more and more accurate predictions and ultimately better decisions.

The formula to obtain the holy grail of Artificial Intelligence and big data analytics is quite simple:

Data Value = mountains of (curated) data + ML algorithms.

In other words, data is valuable if it enables business or societal applications of accurate predictions based on huge data sets and appropriate algorithms.

But this iterative equation demands serious quantities of fuel to run continuously – and that means vast amounts of ubiquitous computing power and data storage. My colleague Johan Torstensson also discussed this in his recent blog, where he notes multi-cloud systems are the new norm, and organizations need to be aware of the silos.

Don’t fight data gravity

We have seen many enterprises make over-simplistic conclusions like this one. “If we want to run any AI on IoT use case, we simply use cloud solutions. Here, we have an IoT business implementation for the predictive maintenance of our oil rigs with intelligent valves in the pipes – let’s just run everything on AWS.”

In at-scale implementations of this kind of IoT initiative, early adopters have learnt that this is a very expensive and often ineffective solution. There are many factors that limit central cloud solutions alone from always working optimally for IoT applications, but three issues stand out:

• Network latency and data gravity – Data gravity is based on the idea that large data sets have mass, like the planets, and the higher the gravity, the more applications will be attracted to where the data is created and used. Applications requests often can only tolerate very low network latency. This can be life critical in certain health IoT scenarios or self-driving vehicles.
• Cloud and connectivity costs – IT economics indicates that data transfer costs and the high cost of central cloud resources constrain implementations. There is a clear need to find the optimal placement of compute and data storage resources to avoid prohibitive cloud costs and increase real time IoT performance.
• Data sovereignty – Global IoT services will face different local regulations that will demand that the data is located within local boundaries.

Public cloud alone is not an optimal engine for the data economy

Cloud services truly power the data economy, but this goes beyond what we know as the public cloud. Cloud resources are genuinely ubiquitous, spanning hybrid and multi-cloud environments and distributed along all points from core cloud data centers to Edge IoT devices and even sensors. Solutions beyond the public cloud present remarkable opportunities for increased efficiency.

If you consider network latency and data gravity, data sovereignty and economic factors together – along with the fact that many end-point devices run on batteries and that data transfer is very energy hungry – you will understand why Edge computing is positioned to blossom. Edge computing is based on the need to find the right balance between the placement of data, computing power to run decision-making algorithms, network design and pervasive security. This complex of models needs to be optimized with ML-aided IoT orchestration platforms.

Core to Edge hybrid and multi-cloud continuum solutions power “liquid IT”

Digital pervasiveness is enabled by digital technology setups that I call “liquid IT”:

• liquid computing – anywhere and optimized computing based on virtualization of network and data center resources,
• liquid data – data-flows and optimized data placement based on type and use of datasets,
• liquid applications – API-integrated microservices running on containers.

The fluidity of users, applications and data powered by APIs, policy-driven governance, automation, etc., needs to be accelerated to truly serve our customers’ business success through digital solutions.

We need to do that with robust end-to-end orchestration. Once we have understood this, we can solve the governance and management of “liquid IT”. This requires some strategy, as shown by the example of a solid ice ball. At first, it is relatively easy to hold in your hand, but when it melts and becomes water, it is impossible to hold, no matter how strong your grip is. This is the problem with “liquid IT” – it is way too complex. You need a “container” for your “liquid IT” to prevent it from spilling .

This “liquid container” is composed of the software-defined platforms that create the “guard-rails” to govern and orchestrate your IT. This can be as simple as a multi-cloud platform deployed and operated as code by the cloud platform/SRE team in a pure-play cloud environment. Alternatively, this can be a unified management platform for integrated management of both your “solid” and “liquid” IT – or should we say… “melting IT” to describe it being in a transition to the cloud.

This management platform is powered by AI/ML and automation (AIOps), enabling end-to-end operation and governance of that “liquid IT”. Traditional ways of managing IT are not designed for digitally intensive IT. Instead, they are designed to manage traditional “solid IT”.

There is no time to lose to deploy the digital fabric of hybrid and multi-cloud infrastructure, that will unlock the value of fluid data and accelerate digital development process in your business without creating multi-silos. My advice: consume as a service, based on pre-configured templates and deployment frameworks.

Connect with me in LinkedIn to continue the discussion or reach out via the contact form.