Secondary menu

You are here

Industry Standard Servers (x86)

Servers come in various shapes and sizes, typically falling into the following categories

Universal or Tower Servers

Tower Server

 

Similar in layout to Tower PCs but typically larger with lots of expansion room internally for disk drives, memory and processors. Sometimes housed under a desk or more often on a shelf in a computer room. They can often be mounted sideways into a 19 inch rack by adding rack mounting rails. (Racks are large metal cabinets used to house multiple servers and other devices in computer rooms and data-centres. 19 inch refers to the standard width catering for devices from multiple vendors.)

Rack Optimised Servers

1U Rack Server

 

These are typically much lower in height than Tower servers – designed to make better use of the space available within 19 inch racks. The height is measured in Units or U – one U being 1.75 inches. The most popular formats are 1U and 2U.

Blade Servers

Blade Chassis

 

These slot into a proprietary blade chassis. The chassis itself is then installed into a standard 19 inch rack. Depending on the blade manufacturer, the chassis can house varying numbers of servers as well as other devices such as storage blades and networking switches. Blades offer the ultimate in rack density and by accommodating multiple devices connected by a neat backplane – cabling and device management can be simplified. All of this comes at a price which means blades are not right for every situation.

Common server roles include the following:
A File Server is used to store and share files: spreadsheets, Word documents, pictures, MP3 files etc. This type of data is referred to as “unstructured data”. If you have a shared network drive which you use at work, this will probably be sitting on a file server. The performance demands on file servers are relatively low and often these servers double up to manage print jobs – temporarily holding items to be printed and coordinating output to networked printers. This dual role leads to the term “file and print”.

Application Server is another generic term and refers to machines which can run a wide range of applications such as warehouse management systems, contact management or order entry systems and can also be applied to all of the following more specific roles.

A Database Server runs a database application (for example Microsoft SQL, Oracle or IBM databases). The type of data here is referred to as “structured” data because within the database it resides in pre-defined tables of information. Databases store, retrieve and manipulate data.

Web Servers have a different role which is to present information to users accessing websites. This could be internal or external to the organisation. Web servers need to support potentially thousands of users.

Some servers are used to run software which “runs” the infrastructure. A common example is a Domain Controller which is a key component within a Microsoft based infrastructure. It controls users and devices and the permissions which enable them to interact. User network login is an example of this in action.

E-mail Servers. The applications here interact behind the scenes with the e-mail software running on your PC or laptop. E-mail has become very critical to most organisations so these servers need to be specified accordingly.

Traditionally, when applications were put into service, an individual server was deployed for each one (or in some cases multiple servers per application). This has led to inefficient use of server resources and expensive, complex deployments. Server Virtualisation is technology which is revolutionising server infrastructure and enabling one physical server to perform multiple roles.

Maximising Business Value

Business Value

Business Value

  • Cost Reduction. When compared to Unix or Mainframe servers - Industry Standard servers are much lower in cost. Despite this, Unix and Mainframe servers still have their place and architecturally do offer some advantages especially for larger organisations. However, it is the inherent cost advantage of Industry Standard servers that has caused their sales to overtake the rest.

    The reasons behind the cost advantage are: A) widely used components, many shared across the PC market which is huge compared to the server market. High volume drives down cost. B) Research and Development costs are spread across the entire PC market meaning the cost per unit is very low. C) Software providers can develop for a single, widely used platform, again reducing per unit costs. D) The standardised nature of these servers means high competition between providers which drives down costs.

 

Watch out for these

 

  • Despite being cost effective to buy, large numbers of servers can become costly to manage. There is a whole range of tools available to help organisations deploy, manage and eventually retire servers and their associated software.

  • Servers can be under utilised - meaning processor power, storage and memory resources are being used wastefully. Consider server virtualisation to make better use of investments in server technology.

  • The cheapest way to deploy storage (from a capital purchase point of view) is to simply put disks into servers. This can actually be false economy because the storage is hard to size correctly to take into account future needs. Storage utilisation in servers has proven to be low - perhaps an average of 30%. Shared or networked storage may be a better solution.

Marketplace - useful links
Technology Vendor Notes
Server Dell Dell PowerEdge branded servers
Server HP HP Proliant brand of servers
Server IBM IBM System X branded servers
Server Oracle Sun Fire branded X86 servers
Server Fujitsu Fujitsu Primergy branded servers

Terminology Made Simple: