Cloud Computing

Cloud Computing is the practice of providing and consuming IT resources, either hardware or software or both, as service. The consumer, called the tenant, does not own the resources, but pays to the service provider as per the usage. Before we delve any further into this, let’s look at the recent history of enterprise computing.

State of computing at beginning of the 21st century

The foundational work

Few significant developments occurred as we transitioned into the 21st century that laid the foundation for the practice of cloud computing that we see today. Let’s look at each of these:

Birth and evolution of GNU/Linux distributions: In October 1992, Linus Torvalds, a Finnish student released version 0.2 of a free operating system kernel called Linux which he wrote as a hobby project. This kernel combined well with the libraries that GNU had been developing since the 1980s in an effort to have a free Unix-like operating system. The combination gave rise to Linux distributions, which are basically versioned and quality tested bundles of Linux kernel wrapped with GNU libraries and distributed as a complete operating system. By the end of the previous century GNU/Linux distributions had established themselves as open source alternatives to proprietary Unix operating systems. RedHat had released version 7.0 of its distribution (pre RedHat Enterprise one) and Debian had released version 2.2. Linux had come of age.

Virtualization

Traditionally there was a one to one correlation between a computer’s hardware and the operating system. There was one computer running one operating system. More than one user or application could use the computer, even parallelly, but they had a view of the same operations system and hardware. This has privacy and security implications. For example, there were providers that offered shared web-hosting services to the general public. They had one computer running an instance of a web server serving websites of multiple clients. This worked well in theory, but all clients had access to the same system, which meant they had access to other client’s data. There were mechanisms in place to isolate clients, but the mechanisms to do that were difficult and error prone.

Virtualization solved this problem by introducing a layer called hypervisor between the hardware and the operating system. This layer allowed for multiple operating system instances to run on the same hardware by providing each of these operating systems with virtualized versions of the hardware. This bundle of virtual hardware running an instance of the operating system is called Virtual Machine. Now each user or application could have their own instance of the operating system, with not shared access to the same resources, which meant better privacy and security. Though virtualization had existed in some form since the mainframe era of the 1960s, it was VMWare - a company that started in 1998 - that brought virtualization to x86 and made it mainstream. Then there were open source hypervisors like Xen and KVM (a part of Linux kernel) which accelerated the adoption. These were assisted by the hardware support for virtualization introduced by both Intel and AMD in the 2000s.

Containerization

Each Virtual Machine runs an instance of the whole operating system, including the kernel and libraries. Applications directly deal only with the libraries and underlying kernel, though important, is transparent to them. Containers bundle the libraries and application(s) together (not the kernel) and can run on any instance of the operating system (called host). As long as the container’s libraries and operating system’s kernel are compatible it will work. Containers rely on the chroot system call which has been available since early Unix days and use Linux Control Groups and Namespaces for isolation. You can think of them as a sophisticated version of FreeBSD chroot jails, that comes with tools for managing the lifecycle of a container.