In 2022, the cloud will be capable of meeting all a company's IT needs: applications, storage, calculations, massive data analysis, networking, cybersecurity... Amazon, Microsoft and Google are all jockeying to offer their services. Which offer should you choose? Loïc Thobois, an expert in the field, helps us to get a clearer picture.
In the almost 20 years that the cloud has existed, we have lost count of the number of services put on the market. So much so that it is difficult to navigate, understand and compare these services.
To make matters worse, different definitions of the cloud have come and gone. Today, the most popular is that of NIST (National Institute of Standards and Technology), an agency of the United States Department of Commerce.
It defines the cloud by five characteristics:
- Services on demand which require tools allowing administration in complete autonomy for the client to manage the life cycle of its resources and an umbrella contract encompassing the conditions of current, but also future, services of the platform avoiding amendments to each architectural modification.
- A dependence on the network : the connection to the services goes through the network. The latter must therefore be particularly monitored both regarding its availability and its quality (bandwidth, latency, etc.)
- A grouping of resources by pool facilitates the implementation of multi-tenant architecture (each customer will have one or more tenants bringing together its resources). This enables the supplier to benefit from economies of scale when purchasing resources and to optimise their operation.
- Manual or automatic elasticity : scaling tools facilitate the adaptation of infrastructures to needs. This is done manually or through indicators associated with trigger thresholds on certain resources (CPU, Memory, request/s, etc.).
- Supervision of the services made available : the supplier monitors and supervises resources in order to offer optimal quality of service and means of control, particularly regarding billing to customers.
What cloud services are available?
With a few rare exceptions, the various suppliers on the cloud computing market have positioned themselves on services that are often linked to their traditional area of activity.
Thus, we observe suppliers specializing in services
- IaaS (Infrastructure as a Service) providing virtual machine, network and storage infrastructures.
- PaaS (Platform as a Service) which will provide components to facilitate the deployment of application architecture via services ready to be integrated by application developers.
- SaaS (Software as a Service) which offer functionally operational turnkey software that can be marketed directly to end users.
Also, among the most popular services with the highest visibility, we will find the big hyperscale platforms. These are the platforms of suppliers who have fully automated and optimised resource lifecycle management, including Microsoft Azure, Amazon AWS, Google Cloud Platform, IBM Cloud, Oracle Cloud, Alibaba Cloud and Huawei Cloud.
The market's cloud offering
Microsoft Azure
For around fifteen years (2010 for Microsoft Azure), Microsoft has offered one of the broadest catalogs of services on the market, covering all service models. Its platform Microsoft Azure enables the deployment of IaaS virtual machines and PaaS platform components, and SaaS services will also be available via Microsoft 365, Microsoft Dynamics, Azure DevOps, Microsoft Intune, etc.
Far from Microsoft's proprietary origins, the platforms are very open to open source solutions (Linux, PHP, MariaDB, etc.) as well as to the services of first-level partners (Citrix, RedHat, SAP, etc.) and second-level partners (Oracle, F5). , CheckPoint…).
Amazon AWS
For almost 15 years (2006 for the first three services), Amazon AWS has been offering a range of services focused primarily on IaaS and PaaS. It is one of the oldest and most technically advanced platforms in the field.
Benefiting from its historic position, it offers a very wide choice of services in its catalogue. It is the most popular solution for companies that have historically chosen to host a large part of their infrastructure in the public cloud, known as the cloud native like Netflix, Airbnb, Uber…
Google Cloud Platform
Google Cloud Platform (GCP) began operations in 2008 with a few beta services, then officially opened in 2011 with a more comprehensive catalog. Without falling significantly behind its main competitors, a decline in investments for several years has harmed the platform. Investments have resumed in recent years with a more targeted offering than these competitors. So the IaaS offer is technically behind, but the data platform is one of the most powerful.
The advantages and disadvantages of each platform
To evaluate the advantages and disadvantages of an IaaS/PaaS platform, several criteria must be taken into account:
- Simplicity of administration enabling simplified handling by operating teams: Microsoft Azure dominates on this criterion with web tools offering an overview of the tenant's resources regardless of the region and without equal among its competitors. GCP is quite accessible where Amazon AWS requires a much greater investment to master its tools and technical expertise.
- The richness of the service catalog : the offers of Microsoft Azure and Amazon AWS dominate on this criterion even if the offers from GCP and IBM still offer the services necessary for most companies.
- The ability to implement a hybrid infrastructure : Microsoft enjoys a strong presence in the IT systems of many companies. This position offers a significant advantage compared to Google and Amazon solutions for which it is mandatory to refer to partner solutions (VMware, specific private cloud integration, etc.). This involves having to manage multiple suppliers to resolve any operational issues.
- Technical capabilities : Amazon AWS dominates the competition regarding the capabilities to control and optimize the platform. This is also why the big platforms favor this supplier. Other providers offer somewhat more rigid services, so the level of optimization is less important or more complex.
- The costs : the different providers are engaged in a fierce battle over the prices of their services. These vary regularly, but we observe a certain homogeneity between Amazon AWS and Microsoft Azure where challengers like GCP and Oracle Cloud are very aggressive.
As is often the case, each offer corresponds to specific needs. The relative importance of the preceding characteristics will guide the choice of a supplier.
*Loïc Thobois
With more than 10 years of experience in network and system administration and architecture, this Supinfo engineer is a Docker and system supervision expert. Since 2001, he has also been a trainer on Microsoft and Cisco technologies.
