Types of Servers

File Servers

A server is a heavy-duty computer designed to be the core of a network. It does not have any special or exotic hardware in it, but it is designed to work continuously for long periods under heavy workloads to control the network and run the network operating system.

When a network is based on a central file server, it allows networked computers to share resources such as printers, and users can store their files in a secure centralised location. The file server offers services to all users, but each workstation on the network does its own processing. For example, if a user wanted to use a spreadsheet she would run the spreadsheet on her own PC, but would be able to save her work in her home directory on the server, and use a shared network printer.

File servers typically have very large hard disks (sometimes using RAID), lots of memory, and are stored under high security. In large networks, there is often more than one file server, to distribute the work load or to handle specialist tasks (such as a proxy server, web server, email server, login server, print server, CD-ROM server etc.)

Since the file server is the heart of the network, if it fails it can severely affect network users, so it is usually treated like royalty. They are usually kept firmly under lock and key in air conditioned rooms with uninterruptible power supplies (UPS) to protect against power blackouts, brownouts and voltage spikes.

Most file servers have large capacity backup devices such as QIC drives so the server can be backed up regularly.

See here for a discussion of file server components.

Proxy Servers

Proxy servers are computers (or software) through which all requests for downloads pass. The proxy server determines how each request is handled.

• If a request has already been downloaded recently and is in the stored cache, the proxy can deliver the file from the cache quickly and with no extra downloading.
• If the request is not in the cache, or it is too old, the proxy will download the file and serve it to the requester.
• If the proxy server has been programmed to forbid certain download locations or file types or protocol (e.g. FTP) , it will reject the request.

Proxy servers can greatly increase surfing speeds and reduce bandwidth costs.

Proxies can also be set up in software applications: your browser, for example, has its own cache to speed up repeated webpage accesses.

If you download a page and you know it's out of date, it's often because you were served a cached copy of the file. When you refresh the page in your browser and it appears up-to-date, you have wiped the obsolete file from the cache and forced a fresh download. Note the difference in times between the obsolete page to load and the fresh page to load.

Wikipedia adds: A proxy server has a large variety of potential purposes, including:

• To keep machines behind it anonymous (mainly for security)
• To speed up access to resources (using caching). Web proxies are commonly used to cache web pages from a web server.
• To apply access policy to network services or content, e.g. to block undesired sites.
• To log / audit usage, i.e. to provide company employee Internet usage reporting.
• To bypass security/ parental controls.
• To scan transmitted content for malware before delivery.
• To scan outbound content, e.g., for data leak protection.
• To circumvent regional restrictions.

Application servers

While a file server (as its name suggests) serves out files to users, an application server actually runs programs on behalf of users. Similar to the earliest mainframe computers that ran programs for users and send the results of the calculations to the users, an application server actually runs application programs like Word, Excel, databases itself.

Why?

It allows Thin Client Computing (e.g. Citrix Metaframe and Windows Terminal Server)- users' workstations need only be very cheap and low powered (really just a keyboard, screen, mouse and enough IQ to send the user's typing and clicking to the application server. The application server is the real brains of the outfit. It runs the programs for the user and sends back screen images that appear on the user's screen.

The benefit for an organisation is that they can use any old nasty, horrible, low powered computer as a workstation and it will behave as if it were a firebreathing, supercharged state-of-the-art PC. This is handy, for example, if the organisation needs lots of workstations but does not have much cash; or if they need to put workstations in high-risk areas such as factories or on construction sites. If the PC is wrecked by dust or is stolen, it does not matter a lot because it was cheap and there is no valuable data stored on it: all data is stored securely and centrally on the application server.

Another benefit is that network managers do not have to configure hundreds of workstations: any change to software or operating system settings is done once, on the application server. It is far easier to install a new program once, on the application server, than it is to go around and install it on 200 workstations!

So why isn't everybody using thin client rather than the traditional high-powered "fat client" workstations?

Naturally, if (say) 200 users are running big programs on a central application server, the server has to be VERY powerful and have HUGE amounts of memory. Application servers typically have twin CPUs, a gigabyte of RAM, massive hard disks... very expensive! Any failure on the server means everyone is offline. Furthermore, multimedia programs don't work well on thin client because they drown the network in traffic (i.e. it's superfat traffic on a thin network).

Both Microsoft and Citrix require a high-end server for acceptable performance. For example, the server should not provide other critical network services and multiple LAN adapter cards are recommended. For each server, Microsoft suggests 30 users for heavy use to 75 users for light use. Besides the baseline memory requirement, Citrix recommends 4 MB to 8 MB for each additional user; Microsoft suggests 10 MB to 21 MB for each additional user.

Still, there are enough advantages for some organisations to swear by thin client and application servers: but they're not for everybody... yet!

Server Blades

A server blade is an entire server (processor, system memory, network connection, and associated electronics) on a single motherboard which slides into an enclosure that can contain dozens of other blades. The enclosure supplies all of the blades with power, fans and cabling.

Unlike traditional servers in individual boxes, the use of server blades takes little physical room and allows thousands of blades to be deployed relatively cheaply. It is also quite easy to service the hardware, increase its power and manage the hardware resources. Installing, servicing, and removing blades is much easier than working with chassis-mounted servers. Shared power supply, cabling, fans and storage reduce the number of duplicated and failure-prone components in the environment. IT managers can easily monitor, configure, and troubleshoot systems. Organisations can also add to their systems in much smaller, more precise increments. Perhaps most important, blades offer the highest computing densities : they can't be beaten when measuring how much processing grunt you can get into a square metre of floor space.

Information Appliances

Information appliances are specialised servers dedicated to one task. They are extremely easy to install and add to infrastructure. Using specialised and proprietary components with embedded operating systems, they require little or no no administration. Their abilities are preinstalled and preconfigured, and web browsers are used for configuration and administration. They are the classic "black box" devices where the user does not need to know how they work, just that they do work.

Information Appliances (also called network appliances) are a growing new class of dedicated devices that include: Internet (web) servers, Proxy/caching servers, Database servers, Email servers, and File servers.

Their benefits include:

• easy and quick installation
• big cost saving for management and administration
• easy remote configuration through a Web browser
• operators do not have to be computer experts
• they need minimal user intervention
• they provide support for multiple versions of every leading operating system and are not tied to a specific OS
• many are upgradeable via FTP or downloading code from the Internet
• they should not require any client software (programs on other computers in the organisation)

A common information appliance is the multipurpose server appliance that provides, in one little box, such features as:

• Internet connectivity
• website hosting
• a proxy server, to cache (store) downloads to avoid having to download them again soon afterwards
• firewall servers
• web access
• e-mail
• FTP
• Virtual Private Networking (VPN) facilities to connect employees, customers and e-business networks
• print server functions (controlling printers on the network)
• file server operations
• enterprise database management
• transaction acceleration software
• cluster administration (managing groups of servers)
• e-commerce hosting (e.g. website "shopping trolleys")

The product usually consists of an embedded OS with features like e-mail and Internet access managed through a browser, network-connected hardware, a hard disk, a processor, routing software, and a modem. Most appliances use a version of Linux running on an Intel processor.

Who uses information appliances rather than traditional servers? Those who want a job done, but don't want to be forever fiddling with settings and software installation. Dedicated information appliances often lack the sheer power of traditional servers, but score points for ease of use, flexible management, and reliability. Small organisations with little IT expertise would find such appliances attractive.

The drawback to such appliances is a result of their "sealed unit" nature: they are not usually designed to be expanded or seriously reconfigured. Like laptop computers, what you buy will be all you will have for the life of the product.

How servers are different to desktop computers

Servers tend to differ from the average computer in terms of:

Memory - servers require large amounts of RAM to work efficiently.  Application servers require huge amounts of RAM to run programs for remote users.

Storage - servers need very large and fast hard disks. Many large servers use RAID (Redundant Array of Independent Disks) arrays for maximum reliability and speed. RAID uses a group of hard disks that work as a unit and usually offer built-in backup of data. RAID disks are usually "Hot Swap" drives that can be removed and inserted without turning off the server or disrupting the network's operation.

Processing power - while processing power is not so important in a file server, application servers benefit greatly from having multiple CPUs.

Backup - most servers have inbuilt high-capacity tape backup drives to protect against data loss. Tape drives usually use QIC (Quarter Inch Cartridge) tapes.

Connectivity - servers often have two or more fast network cards to multiply the rate at which they can send and receive data to the rest of the network.

Robustness - since servers run all day for years on end, the components used in them need to be of higher quality than those in the average desktop computer. As you may suspect a server will cost far more than a desktop PC - perhaps up to ten times as much.

Scalability - servers are designed to be expandable as network demand grows. Scalability is the ability to increase the size and power of equipment and networks as required. Many servers let you add up to 8 hard disks, two power supplies, two NICs, two CPUs etc. The design and engineering of this expandability is expensive.

The larger the network, the more work a server must perform. Bigger networks often use multiple servers to share the workload. Special load balancing software is used to distribute the workload efficiently between the servers.

Sometimes, networks use specialist servers that are dedicated to jobs such as:

• Login servers - which authenticate users
• Proxy servers - which cache recent downloads in case they're requested again
• DHCP servers - which allocate IP addresses to users as they log in
• Print servers - which manage the queues of print jobs
• CD towers and Virtual CD servers - which imitate large CD towers
• Web servers and FTP servers - which handle requests for web pages or files from the Internet
• Email servers - which receive, store and distribute email

On smaller networks, most of these services are performed by software in a single server. Overworked servers can be helped by using server blades. These slim units plug easily into a network and act as file servers. Network storage can be enhanced by plugging in NAS (Network Access Storage) units. With up to 960G of hard disk in a slim package, NAS devices greatly expand a network's storage capacity.