Skip to content

The Progression of Server Hardware

A replica of the NeXT machine used by Tim Berners-Lee, 1990. Image provided by CERNWhat is a server?

The term "server" has a dual meaning. It refers to both a physical computer that provides resources over a network and the software program running on it.

Server (hardware): A physical device connected to a network, hosting one or more software-based servers alongside the operating system. Also known as a host, any computer can serve as a hardware-based server with appropriate server software installed.

Server (software): Server software is responsible for overseeing the hardware and software elements within a network, enabling users to collaborate on files, printers, and additional assets. Numerous server software options exist, each offering unique functionalities and abilities.

lim Burners-Lee at his desk in CERN, 1994. Image provided by CERA1990: First-Ever Online Server

In 1989, Tim Berners-Lee, an English scientist at CERN, developed the World Wide Web to facilitate global information sharing among researchers. He created the world's first web server on a NeXT machine with a 2 GB disk and a 256 MHz processor. The website contained links to information about the World Wide Web project and web server development. By late 1992, the World Wide Web network had expanded to include numerous accessible web servers.



1993: Rack DevelopmentCompaq Rack-Mountable ProLiant brochure, 1994

The emergence of server hardware in 1993 led to the introduction of rack-mounted servers, which utilise a system of mounting slots to accommodate multiple servers in a single rack. This design optimises physical space, allowing companies to consolidate servers into smaller areas. However, this setup can lead to heat accumulation, necessitating sophisticated cooling systems. Consequently, businesses began consolidating their technological infrastructure into dedicated server rooms, addressing temperature control and protection concerns and paving the way for the modern data centre.



RLX Technologies System v Blade server, 20022001: Blade Development

Christopher Hipp and David Kirkeby, in 2000, submitted a patent for a blade server. Their firm, RLX Technologies, launched the first commercially successful version the following year. Blade servers revolutionised server hardware by addressing flaws in rack-mounted servers, offering fewer components, lower power usage, and reduced volume. They are designed to fit into blade enclosures alongside cooling and networking hardware, maximising density within data centres. This innovation significantly improved productivity by enabling more efficient use of computing resources.



2005: Evolution of Server Managementserver technicians at a server farm

With the introduction of blade servers, focus shifted towards improving management for consistency and productivity. Server clusters, for instance, increase uptime by redistributing load if one server fails. Additionally, out-of-band scheduling or remote management came along, allowing IT teams to control servers remotely, enhancing reliability and reducing the need for on-site administrators.


HP Moonshot at Grenoble, 2014

2013: First Software-Defined Server

In 2013, HP Labs introduced Moonshot, the world's first software-defined cloud. Moonshot servers consume less energy and space compared to traditional servers. They are tailored for diverse data centre tasks, including large-scale data processing and high-performance cloud computing.

Around this time, virtualisation gained momentum, enabling the creation of virtual servers, or cloud servers, which replicate the capabilities of physical servers, simplifying server management by reducing physical infrastructure requirements.


2020: Edge Computing & AI Accelerationhands-of-male-it-support-professional-adjusting-ha-2023-11-27-05-05-24-utc

In 2020, the growing demand for edge computing drove the development of servers optimised for edge deployments, bringing processing closer to where data is generated. Additionally, specialised hardware accelerators such as GPUs became increasingly important for AI and machine learning workloads, leading to the development of servers tailored for AI inference and training.


*Info sourced from: