The idea that many users share a physical system stemmed from the scarcity of resources: there were times when processor capacity and memory were scarce and expensive.Thus, services were sold as a service to customers. Large machines(mainframes) performed tasks(workloads)in small increments (in time slices).This computer time could be bought.
There was also a corresponding development in the field of web publishing: You could configure servers so that each user got their own area, a homesitethat could be accessed from the Internet.Since Unix systems (among others) were often used for such systems, a graduated user authorization was available. Only the owner had access to this home area writing, via Internet (www) all. The individual homeareas were separated from each other.Students in particular were able to use the opportunity to set up their own homepage on the university computers. At that time, these sites were still written with Tilde: http://www.uni-irgendwo.edu/homepages/~my-homepage/
Later, the web server software made management more convenient and supported vhostrecords, i.e. “virtual hosts“, which were easily separable from each other (via rights management).Each user had their own webspace, the resources of the machine and the web server were shared (Shared Hosting).
In the course of the further development of the operating systems and the hardware, one could also offer different script interpreters (perl, later PHP).This allowed you to create scripts (programs) that the server was performing. Databases were also available. This also allowed dynamically generated websites to be operated on shared servers. This enabled the provider (Hoster) to offer the services very cheaply, in the form of defined products.
However, this virtualization could be driven even further.Not only could directories be separated and the resources of the web server divided, but other resources could also be virtualized using suitable middelware.In virtual machines, a separate complete operating system environment is provided (complete directory tree of a Linux distribution, for example), so that this server behaves outwardly like a physically existing server.Several virtual machines can be operated on this physical machine at the same time and different operating system variants can be used.
In the case of the cheap virtual servers, the core area is divided, i.e. the Linux core, which cannot be changed in this form of virtual machines.
What is clear is that the performance of this host system must be correspondingly high (many processors, many RAM modules).The virtual machines run in isolation from each other in containers.To divide these resources, you use a special form of software called a hypervisor.
In addition, it is also possible to not only virtualize the software (operating system and applications), but also the hardware can be emulated.This full virtualization requires even more powerful hardware for the hostsystem.
Since you had the individual virtual machines in handy containers, you could easily move these containers from one physical machine to another, copy them, etc.This means that you can set up and start another server with the same software base by clicking on it. This allows you to scale the number of available servers and distribute the incoming load, for example via load balancers.
Other elements of typical IT infrastructure can also be virtually mapped, including network access (interfaces), which can be adapted as required.
In this way, IT services can be provided to customers on demand, whereby the customer can flexibly choose the scope of the service and can also change it at any time.This enables provisioning and billing on an hourly, minute or second basis, depending on the business model.
In itself, these models were not new.Online retailer Amazon has begun to build up computer power on a large scale and to virtualize it for ease of management. Gradually, unused capacity was offered on the free market. Today, Amazon Web Services (AWS) is an important business of the company. Other IT giants followed this and followed with similar business models and supply structures, e.g. Microsoft Azure, Oracle, Google Cloud, Alibaba.