The cloud is like an all-you-can-eat buffet! Unlimited access to dematerialized resources, consumable on demand: storage, databases, applications, computing capacities... and now artificial intelligence. Faced with the multitude of solutions available, choosing can be complex. Private, public or hybrid cloud, multicloud: what are the differences? And how to make the right choice? Our expert, Loïc Caroli, answers you.
The cloud has become omnipresent in our daily lives. The general public knows it through video or music streaming platforms, social networks, office applications or online storage. Professionals use it in their business applications (CRM, business intelligence, project management, HRIS, etc.) or to collaborate with each other (Slack, Teams, etc.).
For their part, IT professionals juggle subscriptions and APIs for different cloud services, trying to monitor their usage, security and cost. The challenges related to its management and security require expertise and constant monitoring.
Although the cloud has existed since the early 2000s, its adoption has experienced several phases of acceleration. The first, in the 2010s, with the arrival of hyperscalers, Amazon Web Services (AWS), Microsoft Azure and Google Cloud Platform (GCP) which have increased innovative services and lowered costs.
The second, during the Covid epidemic where teleworking reinforced the need to access applications and data remotely.
The third is currently taking place, with themeteoric adoption of generative artificial intelligence (GAI).
AGI requires, in fact, a large amount of computing and storage power that the cloud can provide in a flexible and scalable manner.
What is the cloud?
An intangible concept, cloud computing nevertheless has a fairly simple definition. Designates any set of distributed servers that host software, infrastructure and data accessible to users, via an extended network, against a subscription. In short, any IT resource available whenever you need it, dematerialized and for rental.
The 3 cloud services
These resources can take the form of an application, a platform or an infrastructure. Their access to end users is generally via a simple web browser or via APIs.
- SaaS (Software as a Service)
Le SaaS is a software application, web or mobile, provided as a service. This is a rental of a ready-to-use service. Unlike software that is installed locally on a computer or an on-site server, you do not need to install it for each user or perform any maintenance task (bug fixes, updates, etc.) .). SaaS is therefore aimed at end users.
- PasS (Platform as a Service)
Le PaaS adds an additional dimension compared to SaaS: the hardware is added to the application to form a platform. PaaS operators provide a remote development environment, but also middleware. PaaS is therefore aimed at developers.
- IaaS (Infrastructure as a Service)
The model IaaS corresponds to the rental of a technical platform, including servers, applications, databases, etc. IaaS is aimed at teams of production, but also to developers. Certain uses are subject to variations such as DaaS, Desktop as a Service, FaaS, Function as a Service or CaaS, Container as a Service.
The cloud and containers
The cloud is part of the trend of service virtualization by pushing it to a higher level which is the containerization material infrastructure. Storage, memory, computing power, network, etc. are delivered in the form of containers, accessible on demand, via the Internet.
It consists of bringing together the software code and only the components necessary for its execution, libraries, frameworks and other dependencies so as to isolate them in their own “container”. The software in the container can therefore be moved and run consistently across all cloud models, regardless of their operating system. The container functions as a sort of bubble, or computing environment that wraps the application and isolates it from its surroundings. It's actually a complete portable computing environment.
The 5 characteristics of the cloud
If we summarize, the cloud is defined by 5 main characteristics:
- Access from anywhere via the Internet or a corporate network: accessible anytime, from anywhere and from any device. Access is via browser, open APIs or SDKs.
- Self service : access to the service is via a web interface, APIs or SDKs. Delivery of the service takes place in seconds or minutes.
- Dematerialization and sharing of resources : Costs are reduced with this rental model because teams do not have to purchase their own hardware resources.
- Flexibility (or elasticity): resource allocation and scaling of resources on demand is done simply and automatically. Cloud applications can automatically reduce or increase their resources in response to peaks in activity.
- Pay as you go (Pay as you go): users pay for resources according to their consumption. IIt is thus possible to optimize its Capex (investment) for the benefit of its Opex operating budgets.
The 4 cloud models
There are four cloud deployment models.
- The public cloud
The public cloud is operated by a third party which owns the IT infrastructure and makes it available to an end user. Amazon Web Services (AWS), Microsoft Azure, Google Cloud (GCP) and OVH in France are the main public cloud providers. A public cloud is shared among several clients.
- The private cloud
The private cloud includes computing resources dedicated exclusively to an organization. It can be physically hosted internally or externally.
- Hybrid cloud
It is a combination of public and private cloud that operates as a single environment.
- The community cloud
This cloud is shared by several organizations. Generally hosted externally, it can be hosted by one of its members.
And four levels of maturity
- Cloud First
most commonly refers to technology that was typically designed as a server-based technology, but is now being used and applied in the cloud environment. For example, a payroll application that was originally designed for an on-premises server, but was migrated via lift-and-shift.
- Cloud Ready
used to describe applications and technologies designed to run on a cloud computing platform. This technology allows organizations to move their applications to the cloud without making any changes to the source code. This allows businesses to take advantage of cloud services such as scalability, cost savings and increased speed.
- Cloud Native
refers to applications, technologies and software created and deployed in the cloud using the principles of cloud and cloud-native technologies: containers, serverless, APIs, object storage, etc. Because these apps are built from the ground up in and for the cloud with the 12 Cloud Natives factors in mind, they are considered the most compatible and optimized apps for the cloud.
Which cloud to choose?
Choosing a cloud solution depends on your use cases, your technological footprint, your regulatory requirements, but also the maturity of your organization in terms of skills and operating model.
Le public cloud is more suitable for heavy workloads, aow fluctuating demands or the need to subscribe to innovative offers. THE private cloud is best suited to predictable, constant loads and applications that host sensitive data. THE hybrid clouds are more versatile and meet with great success in 2024.
Which is the most secure?
Private clouds are generally considered more secure, because these architectures are not shared like public clouds, which are more exposed to security breaches. More and more companies are abandoning their 100 % public cloud strategy to repatriate their sensitive data to data centers dedicated to them.
Which are the most expensive?
This is a complex question because costs vary according to many parameters such as IaaS, PaaS or SaaS models, the regions where the resources are deployed, the selected suppliers, the Customer's consumption commitment in volumes and duration.
It is essential to implement a FinOps approach starting with a business case.
However, private clouds based on hyperconverged infrastructures and proprietary bases are more expensive in terms of capital expenditures, but less in terms of operating expenses than public clouds.
Don't hesitate to operate multiple cloud architectures simultaneously and orchestrate your workloads dynamically with a cloud management platform!
Depending on how your data is classified, you might run some workloads on one or more public clouds and others on a private or hybrid cloud.
This strategy, although it involves managing a larger number of clouds, offers you unparalleled flexibility, an answer to the risk of dependency compared to the use of a single public, private or hybrid cloud.
