What is at stake in the cloud for a large organisation
A request for cloud expertise rarely arrives at the moment an organisation discovers the subject. It arrives when the cloud is already there: applications migrated opportunistically, teams consuming managed services each in their own way, a bill growing faster than usage, and management wondering whether the original promise has been kept.
Four shifts explain this unease.
Responsibility shifts, it does not disappear. The provider is accountable for the infrastructure it operates. Configuration, access rights, data, backups and service continuity remain your responsibility, to an extent that varies according to the type of service subscribed. When this boundary is poorly understood, entire parts of the system end up without an owner.
Spending changes in nature. It is no longer set once a year in an investment plan. It is decided every day, with each resource deployed by a team. Traditional budgetary control only sees the result when the bill arrives.
Dependency becomes a management issue. Concentrating critical functions with a small number of providers raises questions of applicable law, reversibility and continuity. For financial entities, Regulation (EU) 2022/2554, known as DORA, makes it a fully-fledged risk management matter. For essential and important entities, Directive (EU) 2022/2555, known as NIS2, includes supply chain security among the expected measures.
Operations jobs are changing. Infrastructure is described as code, deployed through automated pipelines and monitored by means of service objectives. Teams trained in server administration must be supported towards these practices, not bypassed.
Our view: an operating model, not a destination
A cloud programme is often judged by the share of the estate migrated. This indicator says where the applications run. It says nothing about how they are operated. An application moved as is, administered by hand and monitored as before combines the price of the cloud with the limits of the old operating mode.
Organisations that make the most of the cloud have a few traits in common.
- They decide application by application. A blanket slogan (“everything in the cloud”, or its opposite) spares the analysis and produces migrations without benefit.
- They standardise through the offering rather than through prohibition. A platform team offers services so convenient that product teams adopt them of their own accord.
- They manage reliability, cost and compliance together. A service has an availability objective, an attributed cost and security rules checked automatically. All three can be read in the same place.
- They prepare the exit from the outset. Knowing how to leave a provider is part of the decision to choose it.
Conversely, four situations recur in programmes that disappoint. Cloud governance is designed as an authorisation desk, and the teams bypass it. Security and continuity are added at the end of the project, at great cost. Tools pile up without anyone being responsible for them. Finally, the upskilling effort is underestimated, even though it is what determines the teams’ autonomy after the programme.
Our cloud expertise, from framing to operations
Our consultants work according to a four-stage approach: Frame, Prove, Industrialise, Operate. Framing means taking inventory, deciding and setting a realistic target. Proving means obtaining a first result on a real, limited scope: a first wave of applications, a first service with reliability objectives, a first cost report. Industrialising means extending what has worked by means of a platform and automated rules. Operating means sustaining it over time, with your teams, to whom our consultants pass on their practices.
Six families of services make up this domain. Each one has a detailed page.
- Preparing and supporting the move of applications. Our cloud migration offering covers the choice of a strategy per application, the landing zone, the waves and reversibility.
- Choosing the right level of trust. The sovereign and hybrid cloud page deals with location, applicable law, SecNumCloud qualification and the distribution of workloads between environments.
- Giving teams a short path to production. With DevOps and platform engineering, our consultants help your teams build the delivery pipeline, infrastructure as code and the internal developer platform.
- Maintaining an agreed level of service. The SRE and observability offering puts in place service level objectives, instrumentation and incident management.
- Linking spending to value. FinOps makes costs visible, attributes them to those who incur them and organises their optimisation.
- Continuing to operate despite the incident. Cyber resilience brings together continuity, business recovery and the management of third-party risks.
These workstreams are not all carried out at once. Their order depends on your starting point, and setting it is the purpose of the framing stage.
Cloud, data, AI and software engineering: linked workstreams
The cloud has no value in itself. It acquires value through what it hosts.
Data platforms and artificial intelligence workloads are among the most resource-intensive uses, and those for which data location weighs most heavily in the choice of a provider. Infrastructure decisions and the decisions taken in our data expertise domain therefore benefit from being examined together, particularly on storage, access governance and processing costs.
The benefit of a migration also depends on the architecture of the applications. An application designed for a single server does not become elastic because it changes host. When the target requires the code to be revisited, the work falls within our software engineering expertise: application modernisation, information system architecture, software quality.
Finally, a cloud programme remains a programme. Its governance, its decisions and the support given to teams obey the same rules as any large-scale transformation.
Where to start?
The first step depends on your starting situation. Three cases come up most often.
You have not yet migrated, or very little. Start with an inventory of the estate and a strategy decision for each application. This work avoids building a landing zone that is oversized or unsuited to the actual constraints.
You are already in the cloud and the situation is getting away from you. Start with a review along three lines: who spends what, which services have a reliability objective and an owner, which dependencies would be blocking if a provider failed. This assessment points to the workstream to open first.
A regulatory requirement or an audit requires you to act. Start by mapping critical activities and the providers on which they depend, then measure the gap between what your plans announce and what your teams actually know how to restore.
In all three cases, we recommend aiming for a first demonstrable result on a limited scope before committing to a full programme. To discuss it, contact our team.