Server hardware on the market
From the early days of Novell NetWare-based file servers to today's client/server oriented market, server hardware has really matured. Now, server hardware is predominantly designed to meet specific application needs.
Nevertheless, three simple factors continue to differentiate a server from other computer systems:
- Performance: Essential to provide acceptable levels of service to multiple users, a server can include performance boosting features such as multiple CPUs, high-performance disk systems, and loads of RAM.
- Scalability: The flexibility to add CPUs, RAM, hard disk storage and more in order to cope with increased load or new applications.
- Reliability and redundancy: If a desktop machine fails, one user is affected. If a server fails, the whole company might be stopped. Components with high MTBF (mean time between failure) figures and redundant components for when they do fail can keep your business up and running.
There are servers that are designed to be used for high-availability, heavy use data centres, servers designed for workgroups or small/medium business, servers designed to maximize the amount of computing power that can be housed in 19in computer racks, and even server clusters to deliver massive computational power.
Rack mounted servers: While pretty much all servers today can be mounted in 19in computer racks, 'rack mounted servers' are designed to use the minimum amount of space possible.
Generally one or two rack units high (there are 42 units in a full rack) they range in capabilities from medium to high performance computing. The thinner, one -rack unit systems (1.75in) usually have very limited space for disk drives and little room for expansion. Many organizations deploy these rack servers in the computer rooms where space is limited, or in co-location premises where space is paid for by the rack-unit. The premium paid for their small size is generally easily made up for the space and/or cost saving.
Blade servers: A Blade server is a computer system on a motherboard that contains one or many CPUs and memory.
The thin design allows more computing power to be packed into a smaller space than typical 1U rack servers. These servers are made up of a blade server chassis, into which server 'blades' are inserted. Usually these blades can be added and removed as required, while the system is still switched on. This makes them well suited to high-availability server systems. The complexity of connectivity and cabling is also reduced for these systems. A premium is paid for these capabilities.
Entry-level/workgroup servers: these are generally housed in what most users would recognize as 'server' boxes. They are not overly large; however; little premium is placed on their compact size. They can generally support significant internal expansion hardware including disk drives.
Large organisations will likely deploy most of these server types to provide their computing infrastructure. Rack and blade servers will be used in major office computer rooms, and co-location installations. Rack and entry-level workgroup servers will be used in remote or satellite offices.
Small and medium organizations will tend toward entry-level and workgroup servers and lower-end rack servers.