As digital transformation has become a necessity to remain competitive, many large companies are turning to the cloud, attracted by the promises of agility, cost reduction and innovation. So why do so many of them struggle to maximise their success?
According to Gartner, 85% of companies will adopt a "cloud-first" strategy by 2025 and according to a McKinsey study, the move to the cloud could increase company profits by 20% over a five-year period. However, this is a value proposition that very often struggles to materialise.
The cloud is often perceived as a miracle solution to solve all technological problems, but this perception is misleading. It is not because a company is equipped with cloud infrastructures that it will suddenly see its performance take off, just as owning a Formula 1 car does not guarantee winning a Grand Prix. The value, and therefore the success of cloud infrastructure depends on how its resources are integrated into the IT system, both technologically and operationally. It's all about integrating the cloud into a transformation project that goes beyond the technical aspect, rethinking the processes around the construction, operation and maintenance of IT systems by relying on the essential and exclusive characteristics of the cloud.
Too many companies migrate their existing applications to the cloud without impacting, or only marginally impacting their IT operational processes and models or reference application architectures. They simply replicate what they were doing before, but in a new environment. This environment may have the flexibility necessary to make this technically possible and functional, but it gives the illusion of success while ultimately little (or no) value is generated.
The trap of transposition without transformation
This conservative approach, consisting of "doing as before but in the cloud", does not make it possible to unleash the full potential of the technology. Worse, it can even lead to higher costs and increased complexity. The cloud is more than a new place to host your files: it is a fundamentally different information model, based on the immediate availability of pooled resources. This immediacy has a non-negligible cost, with monthly use of a cloud resource costing around 300% more than an equivalent traditional dedicated resource. Not relying on this specific feature of the cloud to produce value or drive optimisation is mechanically making the transformation a financial failure.
To understand why some companies fail to maximise the benefits of the cloud, it is essential to return to these five fundamental characteristics as defined by the National Institute of Standards and Technology (NIST), and to examine the human barriers to their optimisation:
- On-demand user access to services. The “self-service model”, which allows a user to use (and return) a resource immediately without any pre-validation tickets or processes, is difficult to implement, as it clashes with generally established practices and carries the fear of "reckless and illegitimate spending". Approaches combining self-service and control exist – for instance, offering customised service catalogues and effective reactive control mechanisms. These require the implementation of new processes and evolving your perception and management of responsibilities around data.
- Broad network access. This relies primarily on high-quality and high-capacity service, ensuring that the network’s bandwidth does not prevent you from meeting present or future needs. This is an essential point to validate, particularly in hybrid or private contexts. It is also necessary to implement fast route and flow opening processes to enable rapid connectivity in compartmentalised environments, so as not to compromise the agility model promised by the cloud's ability to quickly evolve.
- Resource pooling. Resource pooling optimises costs, ownership and operation, and contributes to the agility required by modern projects. It mainly concerns the service provider, cloud provider, SaaS publisher or the IT department when it provides a cloud service. A typical example of an error related to this characteristic occurs during container orchestration for a Kubernetes infrastructure, when poorly managed pooling leads to reduced performance as several applications in a common orchestration compete for the same pool of resources.
- Rapid elasticity offers real-time adaptability, guaranteed budget optimisation and improved quality of service by allowing organisations to adjust the resources they mobilise (and bill) to the actual need, including during exceptional activity phases. This feature is an exclusive benefit of the cloud model, and it is often possible to adapt your existing applications to deliver it, sometimes at a lower cost compared to creating new ones. However, to ensure you generate value through this feature, you must mobilise your application architects, who can thoroughly consider whether a new application is necessary and study the changes your existing applications would need.
- Measured service provides the visibility needed to control an environment that we wish by design to be agile, self-service and rich, but that we also wish to pilot from the IT system up to the application level. Measuring service requires collecting, aggregating, synthesising and presenting usage data, which in many cases can be achieved by deploying freely and easily integrated services and tools. Measurement is the first tool for control.
Finally, the general state of mind in a cloud environment must be one of continuous optimisation and support for innovation. Focus on making things possible, facilitating tasks, accepting changes and considering failures or mistakes as the acquisition of knowledge to do better next time. The cloud is not a technical infrastructure issue, it is a paradigm shift that has launched the success of an entire new digital economy and all those who have known how to transform themselves to exploit the cloud model.
Of course, no one expects a large organisation to be turned upside down overnight. The inertia of a technical, methodological and operational context inherited from many years of IT practice makes the evolution long - for tools, for processes and for humans. But there can be no change of direction without a notion of the direction to take or without the will to turn, through successive iterations. With progressivity adapted to the context, the liner moves and each additional degree carries value.
In a nutshell, the cloud should not be seen as an end in itself, but as a transformation tool. Those who know how to exploit it to reinvent their operations will be the big winners of the digital revolution. The others risk being left behind - despite their brand-new Formula 1 car.