01_0789733684_CH01.

qxd 10/18/05 11:31 AM Page 1

1
1.0—Server Basics
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Terms you’ll need to understand:

✓ Server
✓ Client
✓ Proxy
✓ Firewall
✓ Gateway
✓ Router
✓ Bridge
✓ Remote Access Service (RAS)
✓ Domain Name System (DNS)
✓ Windows Internet Naming Service (WINS)
✓ Dynamic Host Configuration Protocol (DHCP)

Techniques you’ll need to master:

✓ Differentiating between various server types and functions
✓ Knowing which networks use rack-mount or pedestal equip-
ment housings
✓ Understanding the advantages of rack-mount server equip-
ment over consumer computers
✓ Moving equipment racks from one location to another
correctly
✓ Identifying typical components found in server rack-mount
cabinets
✓ Knowing the purpose of a keyboard, video, mouse (KVM)
switch
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 2

2 Chapter
. . . . .1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

✓ Listing which components promote fault tolerance through

redundancy
✓ Recognizing unauthorized intruder and environmental alerts
✓ Listing the three major server types and describing their
general applications
✓ Identifying what types of businesses use general-purpose
servers
✓ Reviewing the types of services best suited for the use of
appliance servers
✓ Describing the three application server models
✓ Defining the two distinct services provided by mail server
applications
✓ Detailing how firewall servers protect a network
✓ Explaining the job of a demilitarized zone
✓ Differentiating between two types of proxy server cache
requests
✓ Describing how and why network personnel use management
services
✓ Identifying why messaging, scheduling, and calendar services
are considered useful
✓ Knowing how collaboration services are used to share busi-
ness information
✓ Defining how a news server can be used to receive, store, and
distribute news articles
✓ Listing the ways in which a web server can be used and
configured
✓ Explaining why database servers are necessary
✓ Describing the main purpose behind the use of a terminal
server
✓ Identifying why DNS servers are necessary
✓ Recognizing what gateway and router servers do, and why
they are important
✓ Identifying the minimum requirements of a bridge server
computer
✓ Detailing the proper use of a File Transfer Protocol (FTP)
server
✓ Tracing the development of the Systems Network Architecture
(SNA) server
✓ Explaining how Network-Attached Storage (NAS) servers
alleviate common slowdowns and service interruptions
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 3

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.0—Server
. . . . . . .Basics
3
. . .

✓ Describing how Storage Area Network (SAN) servers access

their data
✓ Identifying which types of clients are taken care of by RAS
servers
✓ Listing the reasons for file and print servers
✓ Explaining why fax servers are ideal for business organizations
✓ Knowing what WINS servers do
✓ Outlining the advantages for using DHCP
✓ Defining a multi-tiered server structure
✓ Learning when back-end and front-end servers are required
✓ Describing how mid-tier machines are backed up in a larger
company

Introduction
The Server+ Exam Objective 1.3 states that the test taker should know the
basic purpose and function of various types of servers. This chapter discuss-
es server types and their functions:
➤ Servers used as gateways, routers, and bridges

➤ Firewall servers

➤ Proxy servers

➤ Database servers

➤ Client servers

➤ Application servers

➤ Mail and FTP servers

➤ SNA, NAS, SAN, and RAS servers

➤ File and print servers

➤ Fax servers

➤ DNS, WINS, and DHCP servers

➤ Web servers

The test taker should also be able to describe the hardware types (blade
servers, tower servers, and rack-mount servers), including their module clas-
sifications, basic specifications, limitations, and requirements (especially with
regard to power and cooling). Each of these types is also discussed in this
chapter.
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 4

4 Chapter
. . . . .1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Comparing Server Systems

Consumer and commercial computers both exist in the form of desktop or
notebook PC configurations. Many can be used as workstations attached to
a network, but they cannot perform the controlling function of a network
server.
Server computers supporting commercial networks are based on physical
configurations that are different from common consumer units, often
employing multiple processors with disk drive arrays.

Server Cases
Servers are housed in one of two different case types: rack-mounts (rack-
mount chassis) or pedestals. The most popular is the rack-mount chassis
because rack-mount components can be slid in and out of the cabinet easily
for inspection and maintenance.
The Unit, simply designated as U, is a common unit of measurement in rack
systems. Rack-mount cases typically come in 1U, 2U, 4U, or 5U sizes. One
U is 1.75 inches high and is the smallest unit that can be placed into a rack.
A full rack is 42 Us in height. The server case may be any size, but is meas-
ured in multiples of the U size designation.

Remember that a 1U server chassis is 1.75 inches high.

Know how to calculate the space that will be taken up by a server rack component of
multiple Us in height.

Rack-Mount Systems
Differences between servers and desktops include the concept of rack-
mounted components. These racks allow service personnel easy access to
troubleshoot, repair, or replace server components. Pullout rails and easy
access panels facilitate maintenance procedures. They serve in business
client/server environments to limit or eliminate downtime due to mainte-
nance or component replacement.
Most racks have wheels on the bottom for ease of movement, but they are
not designed to be moved with components in them. Be sure to move the
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 5

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.0—Server
. . . . . . .Basics
5
. . .

rack empty and then install (or reinstall) the components later. This prevents
damage to the rack components.

Know the correct procedure for moving a rack-mounted server system.

Typical rack-mount cabinets contain many different server system compo-

nent types. A single cabinet may contain several servers, a backup tape sys-
tem, a high-speed Redundant Array of Independent Disks (RAID), and an
Uninterruptible Power Supply (UPS).
Servers do not tend to include a wide array of peripheral devices. When sev-
eral servers are housed in a single cabinet, typically only one monitor, one
keyboard, and one mouse is used. A KVM switch is used to share these
resources among the various servers in the rack. The KVM switch enables
both administrator and users to operate multiple servers by using the same
keyboard, mouse, and monitor.

Know what a KVM switch does.

Inside a typical rack-mount chassis, some components are similar to those in

desktop system units, whereas others are different. Server systems include
familiar system boards, disk drives, power supplies, and adapter cards.
However, they are less likely to possess individual monitors, keyboards, or
mice because they tend to share these peripherals between several units.
Because other computers rely on servers to get their work done, the biggest
requirement for a server installation is reliability. Therefore, servers use
components and configurations that provide fault tolerance for the system.
This is provided using redundant components, where two components work
in tandem so that if one fails, the other will continue working and no loss of
data or service will be incurred. A typical server chassis will house the
following:
➤ A server board (AT- or ATX-style system board)—Uses multiple
microprocessors (from 2 to 32), large quantities of installed Random
Access Memory (RAM), and built-in sensors.
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 6

6 Chapter
. . . . .1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

➤ Multiple redundant power supplies—Used to provide fault tolerance for the

system’s power and to preclude server shutdown due to a power supply
failure.
➤ Multiple disk drives—Arranged in a RAID disk array, they provide high-
speed fault tolerance for the disk drive subsystem.
➤ Multiple network interface cards (NICs)—Used to provide fault tolerance
for the network connection.

Know the difference between improving the fault tolerance of a server, and improv-
ing the capability of a server.

Some of the special server-related advantages to be aware of include

➤ Hinged front panels that incorporate locks to prevent access to control
buttons and switches
➤ Low-friction rails that allow the chassis to slide in and out of the cabinet
for easy access
➤ Slide-off lids for quick access after the security locks have been opened

➤ Cabinet and chassis access alarms to provide unauthorized intruder and

environmental condition alerts
➤ Multiple cooling fans (cabinet fans, chassis fans, processor fans, and
power supply fans) to prevent overheating

Know the advantages of the various specialty features associated with a server rack.

Pedestal-Mount Systems
The pedestal design is similar in appearance to an extra-wide full-tower
desktop case. However, the inside of the case is designed to incorporate typ-
ical server functions. Pedestal cases are used in smaller networks because
they tend to be less expensive.
Pedestal servers feature integrated Ethernet controllers, redundant hot-swap
power supplies, hot-swap Small Computer System Interface (SCSI) drive
bays, Peripheral Component Interconnect (PCI) hot-plug support, and re-
dundant hot-swap fans. A pedestal server’s durability and availability is
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 7

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.0—Server
. . . . . . .Basics
7
. . .

further enhanced by using RAID controllers within a server platform specif-

ically designed to respond to changing business environments. Pedestal
servers are designed to grow along with the size of the business they support.
They typically support from one to four processors, service up to 10 hard
drives, are equipped with eight PCI slots, and operate PCI hot-plug periph-
erals for growth without interruption.

Blade Server Centers

Enterprise networks often include server types that perform different net-
work functions. In rack-mount server environments, multiple-server chassis
are mounted in a rack enclosure with their support systems. They can be
managed at a single location and interconnected through peripheral port and
control interfaces.
To squeeze more servers into less space, blade servers are used. An independ-
ent server blade is a server, mounted on a card that fits in a special chassis, or
shell, approximately 3U in height. These blades are modular, hot-swappable,
and independent. Modular arrangements provide a computing architecture
that improves operational efficiencies. Blade server environments offer a
convenient alternative to conventional rack systems. Several single-board
servers mount inside an enclosure known as the blade center, where they
plug into a common backplane. The backplane allows devices within the
enclosure to communicate without additional cabling, reducing the required
amount by nearly 90%, and providing shared power and interconnection to
other enclosures within the rack.
Blade enclosures require special interconnect cards that slide into the blade
center and collect network and storage signals from the server blades.
Included within the custom-built chassis is the power and I/O connectivity
circuitry. Each blade typically utilizes one or two processors, 512MB of
RAM, three 20GB hard drives, and one or two network connections.
Blade servers increase the capacity of a standard enterprise data-center rack,
and are best suited for working in Tier 1 service delivery networking envi-
ronments, as a front-end system for email, web hosting, directory services,
firewalls, and network management. Although several manufacturers have
recently released multiprocessor, higher-performance blade servers designed
to handle Tier 2 application server environments, and Tier 3 database oper-
ations, questions remain as to whether blades are a suitable platform for
these more demanding workloads, given their existing performance, heat
generation, and cost-efficiency problems.
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 8

8 Chapter
. . . . .1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Know what purpose each environment blade server is best suited for.

Blade management systems monitor the contents of all enclosures, remem-

ber physical device locations, and reallocate resources as needed, to facilitate
hot-plug operations. In addition, they simultaneously enable multiple
administrators to service multiple servers, at the server facility or remotely.

Server Types
Although all servers perform the basic functions we’ve described so far, in
practice they tend to vary significantly in their physical appearance and
purpose. Servers can be divided into three major types: general-purpose,
appliance, and multi-tier, and defined by the primary applications they are
designed to perform.

Ensure that you understand each server type and its general application.

The multi-tier server category is subdivided into three groups: front-end,

mid-tier, and back-end. These subgroups are defined by the position in the
network structure where they are employed, and by their assigned tasks.
Servers falling into these subdivisions have unique characteristics.

General-Purpose Servers
General-purpose servers are intended to provide a wide variety of services, such
as handling departmental email or providing file, print, and web services that
run on an industry-standard network operating system (NOS).
As the most common server products in use today, general-purpose servers
are typically configured with multiple processors.
They are employed in most small businesses because they can be used for
multiple purposes and are designed to do many functions fairly well. In small
businesses, the server must have redundancy to keep its resources available
without undue cost. Redundancy is the practice of devoting extra hardware
to maintain network resources and is the primary form of fault tolerance
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 9

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.0—Server
. . . . . . .Basics
9
. . .

used in server systems. It is created inside a server chassis with backup power
supplies, hard drives, and network cards, and on a system-wide basis by
adding more servers to the network.
Typical redundant items used in general-purpose servers include multiple
hot-swap power supplies, multiple hot-swappable hard drives, and special-
ized hot-swap PCI slots.
Although these items do not individually prevent the server from failing,
they allow components to be exchanged with the system still running.
Combined with the use of redundant components, the effects of hardware
failures are minimized or eliminated entirely.
Powerful processors are selected for use in general-purpose servers to per-
form multiple functions. Such servers utilize two or more Pentium III, Xeon,
Pentium 4, or Itanium processors from the Intel group.
The amount of installed RAM in a general-purpose server is medium to
high, depending on the server’s intended use. Hard drive configuration
depends on whether the client will use it as a file server or not. If so, it will
usually include a RAID 5 array for redundancy, speed, and capacity.

Appliance Servers
Appliance servers provide a single service, such as web serving, or multi-
services such as providing Internet caching and firewall protection. This
server is usually treated as a Field-Replaceable Unit (FRU), so that when it
crashes, it is simply replaced by a comparable unit as quickly as possible. No
attempt is made to repair the unit while it is still connected to the network.
When a company installs an appliance server, two of them will normally be
placed in the network together. If one server fails, the other will guarantee
that network resources remain up and running.
Appliance servers tend to employ middle-level processors and various other
hardware. If the appliance server was intended to provide web services, large
quantities of RAM would be required to enable the caching of whole web
pages into memory. For use as a file server, it would include an advanced
RAID controller to provide improved disk access times, and the amount of
installed RAM wouldn’t be a critical consideration. Appliance servers come
as specifically preconfigured units and are simply connected to the network
and turned on to perform the prescribed function.
Appliance servers are generally in the middle range in terms of cost, and tend
to be less expensive than general-purpose servers. They are typically found
in medium-sized businesses, where using specialized servers to provide spe-
cific functions can be justified.
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 10

10 Chapter
. . . . .1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Application Servers and Services

Server+ Exam Objective 1.4 states that the test taker should know the func-
tion of dedicated application servers, distributed application servers, and
peer-to-peer application servers:
➤ A dedicated application is reserved for a specific need, such as a dedicated
channel (leased line). A dedicated server is a single PC set up in a reserved
network to serve the needs of the PCs within that network. A dedicated
server in some networks will manage communications, and in others it
may manage printer resources.
➤ A distributed application is a single application or group of applications
designed for specified end users. For example, email providers use a dis-
tribution server to distribute the email data to end users who participate
in their mail service.
➤ A peer-to-peer application is the sharing of computer goods and services by
directly exchanging information between users. For example, Napster is
a peer-to-peer server that enables an Internet user to share files with
other users and exchange information stored on their computers. The
resources and services that are generally exchanged include information,
data cycles, and storage for files on disk.

Application servers are configured to run programs that can be accessed by

multiple users, and can be implemented for handling large information data-
bases. Also called an appserver, these machines manage application programs
running between client computers and the organization’s database.
Among the many types of application servers or services in use today are:
➤ Mail servers

➤ Firewall servers

➤ Proxy servers

➤ Management services

➤ Messaging services

➤ Scheduling and calendar services

➤ Collaboration services

➤ News servers

➤ Web servers

➤ Database servers
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 11

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.0—Server
. . . . . . .Basics
11
. . .

➤ Terminal servers

➤ Domain Name System servers

The following list describes the ways in which flow performances are
improved when application servers are used to deliver information:
➤ Security algorithms are applied to both the data and all user traffic.

➤ Client programs are reduced in size and complexity.

➤ The flow of data is cached, and effectively controlled.

Mail Servers
Mail servers send and receive email. These application servers receive and
store electronic mail messages in private mailboxes. seen as folders on the
network mail server. When a user invokes his or her mail client software, the
main mail folder is queried, and the mail server sorts through the account
folders retrieving the mail messages that belong exclusively to the user. The
mailbox processing is performed on the server side, and the results are then
passed to the client.
Mail server applications are actually two distinct services, receiving incoming
email and distributing outgoing email. The normal protocols used are Post
Office Protocol (POP3) for receiving incoming email, and Simple Mail
Transfer Protocol (SMTP) for sending outgoing email.

Firewall Servers
Firewall servers control the connections between two networks, commonly
acting as gateways to the Internet. They protect the network by implement-
ing access control to block unwanted traffic, while allowing acceptable com-
munications. Firewall servers come in both the hardware and software variety.
Hardware firewalls are suitable for co-location duties and for use with parti-
tioned networking environments using demilitarized zones (DMZs). When
a hardware firewall is said to be performing co-location duties, this means
that the firewall resides in a rack belonging to a service provider that may
house other firewalls or equipment used by other companies. A DMZ is a
useful safety feature when hosting a server, in order to separate the corporate
Local Area Network (LAN) from any simultaneous public server activity. It
permits the separation of the administrative and main website traffic.
Although software firewall products may be provided for various server
platforms, some are designed to protect mission-critical applications and
data hosted on Windows servers. They employ intrusion prevention
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 12

12 Chapter
. . . . .1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

systems capable of stopping active and/or date-oriented attacks, and also use
application-hardening techniques. They are designed to be easily deployed,
and to integrate seamlessly into various network infrastructures.

Proxy Servers
Proxy servers are similar to firewalls, and they act as an intermediary between
a client’s workstation and the Internet. Over time, they cache frequently
visited websites. This saves Internet bandwidth and increases access speeds
by allowing the internal network to access these sites from the proxy server.
This type of arrangement works well only for Internet downloading chores.
When both uploading and downloading activities are required, a firewall
server will be used.
Anonymous proxy servers conceal the Internet Protocol (IP) addresses of users
on the network side of the Internet connection. This shields clients from
having their personal IP addresses recorded by hackers trying to gain unau-
thorized Internet access to their computers. Client IP addresses are secure,
as are their browsing interests. Anonymous proxy servers don’t reveal that
clients are using them to browse the Internet, and are commonly used for
web-based email, web chat rooms, and FTP transfers.
Websites can gather personal information about visitors through their
unique IP addresses. This activity is actually a form of spying, and can reveal
personal reading interests. Revealed personal information can make that per-
son a target for marketing and advertising agencies, through the use of
unwanted email or spam. Although proxy servers can help to prevent this,
they can also be used by employers to filter outgoing web requests, or to pre-
vent company employees from accessing a specific set of websites.
Remember that a proxy server is capable of two types of cache requests:
➤ Reverse proxy cache—The cache is configured to act as the original server
from which the client requests the data.
➤ Forward proxy cache—The client requests the Internet data and this
request is forwarded through the cache on the way to the original server.

If the cache does not currently contain the requested information, the proxy
server requests the information from the original server, and stores a copy in
its cache for any future requests. If the cache already contains the requested
data, it is sent to the client without contacting the original server.

Management Services
Management services are special tools and protocols used for remote system
management. The Simple Network Management Protocol (SNMP) is the
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 13

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.0—Server
. . . . . . .Basics
13
. . .

Internet standard protocol to manage IP network servers, workstations,

routers, switches, and hubs. Such protocols enable network administrators to
handle network problems, manage network performance, and plan for net-
work growth. They can automatically notify a system administrator when
specific changes and/or circumstances affect the server.

Messaging Services
Messaging services provide the capability to send instant messages throughout
the network and are usually included in the operating system. The Windows,
Unix, and Linux operating systems include their own messaging service
utilities.

Scheduling and Calendar Services

Scheduling and calendar services help to organize a company’s schedule, or to
document various departmental meetings. These products range from basic
calendars that list important events, to programs that notify employees via
email, pager, or cell phone of group meetings or report deadlines. They also
permit the tracking of product timelines and workgroup projects.

Collaboration Services
Collaboration services allow multiple users to communicate using text, graph-
ics, video, or voice. Whiteboard applications allow the client to communicate
in a text chat box, and draw diagrams to clarify subjects discussed.
Email messages are sent and received through what is commonly referred to
as a client device. Client devices can include personal computers, company
workstations, mobile phones, and Pocket PCs. Centralized computer systems
are made up of servers or mainframe computers, where the enterprise email-
boxes are stored as part of the business network. The email client typically
connects to a network of centralized email servers, which in turn connects to
both the Internet, and any number of private networks. What makes a col-
laboration tool so useful is its family of protective controls that make sending
and receiving electronic information a secure experience for company-wide
operations.

News Servers
A news server is a computer with software installed to receive, store, and dis-
tribute news articles to and from newsgroups on a specific network, or on the
Internet. The most recognizable example of a news server is USENET, an
Internet-based network using the Network News Transfer Protocol (NNTP).
USENET consists of numerous global bulletin boards divided into a variety
of discussion group categories, called newsgroups.
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 14

14 Chapter
. . . . .1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

To connect to a news server, a user needs to know the appropriate server

name, and the port number to be accessed, normally 119. In the corporate
environment, a news server is an extremely useful tool because various com-
pany branch-specific bulletin boards can be organized to focus on a variety
of workplace topics. This enables company employees to pool their resources
and share ideas.

Web Servers
A web server hosts web pages for intranet and/or Internet access, and can host
more than one site depending on its underlying operating system. Web
servers are used to share information and often replace old-fashioned file and
print servers. Today we access websites using browser software, including
Microsoft Internet Explorer and Netscape Communicator.
When configuring a web server with Microsoft Windows 2000, security can
be applied by using
➤ User authentication—The ability to control who can access data on a
website, by forcing a user logon.
➤ Data encryption—The scrambling of data so that it will not transfer
across the network in a readable format.
➤ Web permissions—The granting of access to resources by assigning specif-
ic permission levels to the data, or the folders in which the data resides.

During a browse, the web application server links a database server with the
client. The application server determines if the required information is
already in the database. If so, it is quickly located and transferred to the client
in much less time than it would take to reload the specified web page from
scratch.

Database Servers
Database servers store and sort through data in response to client queries.
Servers of this type are necessary for organizations that must manage large
quantities of data. All data processing is performed on the server using serv-
er-side application software. The client’s software defines the query and is
called the client-side application.

Terminal Servers
Terminal servers are special-purpose computers fitted with multi-ported asyn-
chronous modem connections, and a number of LAN ports. A terminal
server allows serial line devices to connect to the network, with basic
arrangements permitting access only to dumb terminals via modems and
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 15

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.0—Server
. . . . . . .Basics
15
. . .

telephone lines. The terminal server answers modem-based phone calls and
completes the connections to the appropriate node(s). Sophisticated termi-
nal servers permit intelligent terminals such as PCs to connect to the net-
work using higher-level protocols such as Point-to-Point Protocol (PPP). If
network connections are solely obtained through the use of modems, the ter-
minal server is referred to as a modem server.
At a minimum, the terminal server serves these functions:
➤ A remote control solution for remote users

➤ A multi-session operating system, similar to Unix, which allows multiple

user sessions
➤ A centralized management of the networked environment

➤ A solution for lowering the network’s total cost of ownership (TCO)

When connected to the Internet, terminal servers provide either PPP or

Serial Line IP (SLIP) services, and terminals connect to the LAN through
one network connection. The terminal server performs host Telnet services to
all terminals connected to the network, becoming, in effect, a Telnet server.
Part of the Transmission Control Protocol/Internet Protocol (TCP/IP) suite
of protocols is the Telnet protocol. It enables network clients to connect to
Telnet servers and run character-based applications from any remote loca-
tion. In Unix systems, Telnet has historically been the primary method of
enabling clients to access Unix servers and run various applications. In
Windows systems, the primary use of the Telnet utility is by administrators,
to facilitate remote access and remote administration from Telnet clients.

Domain Name System Servers

A Domain Name System (DNS) server contains a database listing, called the
DNS database, used to resolve computer names to IP addresses. These
servers contain mappings between TCP/IP names such as [Link],
and the IP addresses they represent, such as [Link]. DNS caching short-
ens the time taken to perform name resolution and decreases traffic on the
network. Whenever a DNS server receives address information for another
host, or domain, it stores it for a limited time in order to service similar DNS
requests. These repeated requests are not forwarded across the Internet
because the DNS server already has the required information.

Know what type of server resolves TCP/IP names into IP addresses.

01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 16

16 Chapter
. . . . .1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

People deal better with alphanumeric names than they do with numbered
sequences. Conversely, to a digital computer, everything is a number.
TCP/IP networks, such as the Internet, can use human-readable host names
corresponding to unique numbers called IP addresses that have been
assigned to computers. The IP addresses being resolved are usually of the
static variety. For residential Internet users, websites are reached using
modem, DSL, or cable connections through dynamic IP addressing. The
local Internet Service Provider (ISP) temporarily assigns the client an IP
address used only for the duration of that specific connection. A different IP
address will be assigned for the next Internet connection. DSL or cable
Internet services allow the use of a static IP address. A dynamic IP operation
does not affect the typical Internet user because the flow of information is all
incoming. It does present a problem for any client hosting a personal web-
site. A web server using a dynamic IP will be impossible to link to, because
there is no direct relationship between it and its domain name!
Dynamic DNS (DDNS) solves this dilemma through the use of special
DDNS service providers. DDNS is a method of keeping a domain name
linked to a changing, or dynamic, IP address. It allows dynamic IP Internet
addresses to be able to use applications that require static IP addresses by
mapping a third-level domain name to the client’s DDNS IP address. When
the client’s ISP-assigned IP address changes, the DDNS client program will
forward the new IP address to the DDNS.

Know what types of servers are capable of caching and forwarding requests.

Servers Used As Gateways

A gateway interfaces between different types of networks, or protocols. It’s a
special-purpose device that performs conversions between various protocols,
at the application layer. Either a hardware or a software arrangement trans-
lates between these dissimilar protocols. In an enterprise environment, gate-
way servers route traffic from individual PCs to the network serving the
specified web page. They also act as proxy servers and a firewalls. A residen-
tial gateway acts as an ISP, connecting the client to the Internet.
Gateways can exist at the intersection of two networks, such as a business
LAN and the Internet. The end result is the successful transfer of data
between them. Gateways close the gap between two otherwise incompatible
applications, or networks, so that data can be successfully transferred.
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 17

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.0—Server
. . . . . . .Basics
17
. . .

Large corporate gateways can perform translations between internal, propri-

etary email formats and the Internet email format. Often the gateway
conceals the IP address of the sending client, while those accessing the infor-
mation from outside the network only see the gateway’s IP address.
TCP/IP host gateways can access two or more different Ethernet networks,
forwarding messages between them, whereas other hosts cannot. These hosts
are fitted with multiple IP addresses, one for each network being accessed.
Such a gateway might exist at addresses [Link] and [Link], as shown in
Figure 1.1. Host computers forwarding messages between these networks
would first address them to their local gateway. The gateway would then pass
them on to the other network. If a server at address [Link] had to com-
municate with a host at address [Link], it would first have to contact the
gateway machine on its local network. When contacted, the local gateway
would forward the message between networks 192.0.1 and 196.0.1.

Email Server Email Server Email Server Email Server

LAN LAN
Email Gateway
[Link]
Email Server Firewall [Link] Firewall Email Server
[Link] [Link]

Email Server Email Server Email Server Email Server

Figure 1.1 A gateway for network-to-network messaging.

Servers Used As Routers

Not long ago, using a server as a router was a very bad idea, and provided
hackers with unfettered opportunities for compromising a system’s security.
Newer products now secure combined server/router operations by keeping
the routing chores separated from other critical server duties. Terminal router
cards plug into the server and provide the connectivity, performance, and
expandability of a selectable four-port RS-232 terminal server/TBOS port, or
a 10/100BASE-TX Ethernet switch with dedicated routing functionality.
Servers used as routers manage the shared resources of all other routers in the
network, including the various transmission speeds and protocols used.
Routing advertising packets are sent every three seconds or so when the pro-
tocol process is first started, and then about every 10 minutes thereafter.
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 18

18 Chapter
. . . . .1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Servers Used As Bridges

A server bridge requires at least two NICs to connect two groups of comput-
ers. The bridge will act as a server by isolating MAC addresses, managing
network traffic, and translating from one protocol to another. It reduces
network traffic by determining whether a data frame should remain in its
originating network or be forwarded to another.
A bridge contains a list of which computers are on each side, and passes only
packets that must transfer from one side to the other. Network activity is sig-
nificantly reduced when individual groups have heavy internal traffic, but
much less between each other. Computer servers being used as bridges
should have their server software set up first. Then the bridge software
should be installed and configured. Windows XP can be configured as a soft-
ware bridge to permit two or more networks to be tied together to act like a
single network. Software-based bridging requires that the server computer
operating the software bridge be turned on in order for the networks to
communicate.

FTP Servers
An FTP server transfers files across the Internet, an extranet, or an intranet.
To access an FTP server, the client computer must have an FTP client pro-
gram installed. Modern web browsers include a minimal FTP client that
allows files to be sent and received using the FTP protocol.
Many companies utilize an FTP server for software downloads, such as free,
beta, or demo versions, or patches and upgrades. By segregating software
download operations from a company’s web server, the overall level of net-
work traffic is reduced.
FTP is also a viable alternative to using email for transferring extremely large
files. Experienced email users know that large files included as attachments
take considerably longer to send or download with an email server than when
transferred to or from an FTP server.
Most users think about FTP from the perspective of uploading or down-
loading operations, rather than the way in which an FTP server is set up.
FTP software programs are intended to reside on the FTP server itself, and
permit the site operator to control settings that dictate how the FTP server
will operate. FTP server software is used to set limitations on transfer speeds
and maximum users allowed. In addition, specific security measures can be
adjusted, for blocking bounce attacks, hammering, and FXP (server-to-
server copying, or FXP mode, which is built into FTP).
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 19

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.0—Server
. . . . . . .Basics
19
. . .

In a bounce attack, the hacker misuses the PORT command on an FTP server to
indirectly access arbitrary ports on other machines, or systems, not directly
accessible by other means. When the hacker is connected, he or she is free
to bypass any access controls that would otherwise apply.
Hammering is the repeated attempt to connect with an unavailable FTP
server without permitting a normal delay time between attempts. FTP
servers cannot process an unlimited number of requests, and when running
at full capacity, they deny further access until capacity is freed. Hammering
causes a server that is already working at capacity to send a busy response to
any device trying to connect every time a connection is attempted, further
depleting server resources. FTP sites normally require their clients to set
retry times at specific intervals, commonly at least 120 seconds between each
attempt. They also monitor for hammering devices, banning access to
offending IP addresses temporarily, or permanently.
Properly used, FXP is a useful function. However, hackers with bad inten-
tions can also misuse it. An FTP client connects to two servers at the same
time, with one server directed to receive an incoming file, and the other
instructed to send it. The sending and receiving servers connect to each
other, and transfer the requested file without it first being copied to the orig-
inating FTP client. Because this function can cause great damage if improp-
erly used (bounce attacks), it is normally disabled on most FTP servers. FTP
administrators usually enable FXP functions for specific reasons, at specific
times, for specific customers.

SNA Servers
SNA was developed in 1970 as a proprietary protocol for telecommunica-
tions networks, primarily because large corporations demanded more
throughput and better security from their existing networks.
An SNA server allows client access to mainframe (IBM) and mid-range data
facilities, and also permits print facilities to be located on the host computer.
SNA server load-balanced configurations work well with high-priority situ-
ations because the failure of any one server is transparent to a connected
client machine. When one server fails, the client is automatically switched to
the next available server.
Microsoft SNA Server was developed in order to share IBM data in the PC
environment, resulting in seamless mainframe-to-desktop file transfers.
Further refinements in this type of system interoperability resulted in the
introduction of Host Integration Server (HIS) 2000, which has succeeded
SNA Server.
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 20

20 Chapter
. . . . .1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

NAS Servers
Because of slowdowns and service interruptions that commonly occur in
conventional file servers, Network-Attached Storage (NAS) is one alterna-
tive. NAS servers effectively move storage out from behind the file server, and
put it directly on the transport network. Whereas file servers normally use
SCSI and LAN adapters, an NAS appliance uses an NIC to transmit both
LAN and storage communications. Because NAS servers operate independ-
ently from the file server, any client with access rights, anywhere on the net-
work, can directly access stored NAS data.
NAS servers are also optimized for faster processing of I/O storage transac-
tions, avoiding inherent delays from file server hardware or operating
systems.
NAS server arrays can be housed within a 3U rack-mount chassis, equipped
with a redundant power supply featuring an audible alarm. They often use
Serial ATA (SATA) hard drives specifically designed for enterprise operations
featuring 100% duty cycles. NAS server rack-mounts can be fitted with PCI
RAID cards, with each running on its own dedicated PCI bus. Each unit
includes two 2.8GHz CPUs and 4GB of RAM. A dedicated internal hard
drive keeps the operating system separated from the RAID operating system.
Each chassis also comes with an internal CD-ROM drive. These arrays can
be centralized within racks to form systems that are considerably more
robust, as shown in Figure 1.2.
Common rack-mount features include improved manageability, high-end
performance at low acquisition cost, and up to 56TB of data storage per cab-
inet. Easy-to-use GUI or command-line interfaces provide for rapid system
deployment and maintenance. Cross-platform file support is also provided
for any combination of Linux, Unix, Windows, or MacOS environments,
and compatibility exists for leading data backup and replication software.
When a rack-mount system is connected to smart UPS battery backup
equipment, a controlled automatic shutdown can be performed during any
power failure.
To process requests from networks running multiple types of operating sys-
tems, NAS servers and appliances use a common file server access protocol.
The physical complexity associated with using parallel SCSI buses to connect
storage disks to file servers is alleviated.
The main disadvantage of an NAS server is that it shifts storage transactions
from parallel SCSI connections to the production network, forcing the local
area network to handle its normal end-user traffic plus storage disk opera-
tions, including backup. Although file server backups have always resulted in
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 21

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.0—Server
. . . . . . .Basics
21
. . .

large bandwidth consumption, NAS does not resolve this issue by using the
production network for backup and recovery. However, an alternate file-han-
dling strategy called a SAN does remove backup traffic from the LAN.

Figure 1.2 An NAS server rack system.

Although comparisons between network-attached storage and a storage-area

network are inevitable, each type of system has its proper use. Remember
that as the name NAS implies, its data storage can be accessed via a network,
using the TCP/IP protocol. The NAS system is usually configured with
multiple hard drives, and uses RAID for redundancy. One important advan-
tage of NAS servers over SAN servers involves improved scalability.

SAN Servers
A SAN is a network designed to attach such computer storage devices as disk
array controllers and tape libraries to servers. SAN servers are common in
enterprise storage, which focuses not only on storage, but also on data pro-
tection and retrieval within large-scale environments.
SAN servers access their data using low-level block storage methods, as
opposed to the file storage access method, similar to the way in which data is
identified on ATA and SCSI internal disk drives. To access network data,
most SANs use the SCSI communications protocol, without the lower-level
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 22

22 Chapter
. . . . .1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

physical interface structure. Over a shared network, a SAN permits many

computers to access many storage devices.
SAN server racks combine the SAN fabric with RAID disk redundancy to
create an extremely robust and high-performing file system. Multiple com-
puters with different operating systems are enabled for direct access to the
shared file system. If any server connected to the SAN fails, all data is still
accessible by all other servers. In addition, all systems connected to this SAN
server have simultaneous access to the same files, at local disk speeds.

RAS Servers
A Remote Access Service (RAS) allows the client to dial in to a computer
from a remote site. Therefore, an RAS server is devoted to taking care of
these clients, even though they are not connected to the LAN but do require
remote access to it. If a home-based office PC is connected to the main office
LAN via an RAS server’s modem port, the RAS server enables this client to
access files and/or printer services on the targeted LAN. It provides connec-
tivity into a private or corporate network for the remote modem user, neces-
sitating the running of the TCP/IP network protocol encapsulated inside
PPP. The remote machine has access to the targeted network as though it
were directly plugged into it.
For example, when a client dials into the office network from his or her
home PC, the analog modem will dial into an RAS server at the office, and
the client will then be authenticated and granted access. Following this iden-
tification process, printers and shared drives can be accessed just as if the
client were actually at the office and connected directly to the network. In
addition, the server could be configured to only allow access during certain
periods for a particular group of users.
Remember that an RAS server can also be attacked in a similar fashion as
described with FTP servers.
When a suitable RAS server becomes the target, there are techniques hackers
can use to try to break in, such as using common usernames and passwords
like “Administrator” or “root.” To secure the server from this type of attack,
a callback system can be implemented, where even if a would-be hacker is
armed with the correct username and password, the callback goes only to an
authorized client.
For sales or technical personnel out on the road, preventing a hack becomes
harder. In these cases, a token-based security system can be implemented,
where clients log in by using a PIN and a digital “token.”
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 23

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.0—Server
. . . . . . .Basics
23
. . .

Because the token changes at least once every minute, only the token gener-
ator and the server will know what it is.
Callback systems are also implemented for the purpose of reverse charging,
so that the commercial site picks up the employee’s connection charges.

File and Print Servers

The print server decreases the administrative and management workload by
streamlining both local and remote printer control. A client’s print job is
spooled much more quickly by a print server than by sending it directly to a
traditional printer. When using the print server, a client does not have to wait
for the print job to finish before continuing other work. A print server is
often a small box equipped with at least two connectors: one for a printer,
and another that attaches directly to the network cabling. Other print servers
can be equipped with two, three, or four printers, all being operated simul-
taneously.
A file server helps manage access to data files and other network-related
applications. It’s a computer dedicated to specifically storing files, and en-
ables any network client to store files on it. When a system administrator
plans to use disk space to store, manage, and share files and network-
accessible applications, the system should be equipped with a file server.
In large client/server environments, files are stored on centralized, high-
speed file servers that are accessible to client PCs, making network access
speeds much faster than those found on peer-to-peer networks. In order to
permit the tracking of networking tasks, network services such as printing
and email transfers are routed through the file server. Not only can users’
activities be closely monitored, but also inefficient network segments can be
reconfigured to make them faster. Any messages from one client to another
are sent to the file server first. From the file server, they are then routed to
their destination. When dealing with networks comprising tens or hundreds
of client PCs, a file server makes the management of complex and simulta-
neous operations of large networks possible.

Fax Servers
Fax servers help to reduce the amount of telephone infrastructure, while pro-
viding the ability to quickly transport required documents. The authenticity
of printed faxes as legal and official documents, as well as their use for infor-
mation exchange and data confirmation within the corporate environment is
well established. Although fax servers are often standalone operations, they
are sometimes combined with email services.
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 24

24 Chapter
. . . . .1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Most fax server products are software-based, where the administrator is free
to choose the accompanying hardware, such as a network-ready server, a fax
modem, an Integrated Services Digital Network (ISDN) adapter, or a dedi-
cated fax board.

WINS Servers
The Windows Internet Naming Service (WINS) server is a Microsoft
NetBIOS name server that permits the client to search for resources by com-
puter name, instead of by IP address. It provides a distributed database for
registering and prompting for dynamic mappings of NetBIOS names over
TCP/IP for networked groups. WINS maps NetBIOS names by their actual
IP addresses, providing NetBIOS name resolution in routed environments.
NetBIOS names were used by earlier versions of Windows to locate, iden-
tify, register, and resolve names for shared or grouped network resources.
Although the NetBIOS naming protocol established networking services in
earlier Microsoft OSs, it is also used with network protocols other than
TCP/IP. WINS was designed specifically for use over TCP/IP-based net-
works in order to simplify the management of NetBIOS namespace.
The order of the events for a typical WINS operation is as follows:
1. The WINS client registers any of its local NetBIOS names with its
configured WINS server.
2. Another WINS client prompts the WINS server to locate the IP
address for the first WINS client on the network.
3. The WINS server replies with the IP address for the first WINS
client ([Link] for example).

WINS eliminates the need to use local IP broadcasts for NetBIOS name res-
olution, permitting users to locate remote systems on the network more
easily. When a client first joins the network, the WINS registrations are
done automatically and the database is automatically updated. When a
DHCP server issues a new or changed IP address to a WINS-enabled client
computer, client WINS information is updated without requiring clients or
administrators to make any manual changes.
When managing a name-to-address database, or mapping NetBIOS names
to IP addresses, a WINS server is required. Microsoft Windows 2000,
Windows XP, and Windows Server 2003 are configured with WINS server
addresses either manually, or automatically (DHCP), for name resolution.
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 25

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.0—Server
. . . . . . .Basics
25
. . .

DHCP Servers
The Dynamic Host Configuration Protocol (DHCP) temporarily assigns dynam-
ic IP addresses to network workstations and Internet clients each time they
are detected. When these units power down, or disconnect, their IP addresses
become available for reassignment to another client. A group of dynamic IP
addresses, called a scope, will be maintained at the DHCP server. The scope
must be identified during the DHCP server’s configuration procedure.
Temporarily assigned IP addresses from DHCP permit the efficient manage-
ment of the entire IP addressing scheme. Software tracks the IP addresses of
additional computers rather than requiring an administrator to manually
assign them.
Dynamic addressing provides each networked device with a different IP
address every time it connects to the network, sometimes changing them
during a connected session. DHCP also supports a mix of static and dynamic
IP addresses, and Internet Service Providers (ISPs) often reserve dynamic IP
addressing for dial-up users. When the DHCP server assigns IP addresses to
clients using DHCP, it automatically uses leased IP addresses. A standard
DHCP lease is the total amount of time permitted by that server for a client
to use one of its IP addresses, and the DHCP server normally permits the
network administrator to set its lease time.

Multi-Tiered Server Architectures

In large organizations with complex networks, the concept of using special-
ized servers to perform specific functions is further refined by organizing
groups of servers into a tiered structure. These tiers optimize the use of dif-
ferent server types for performing related functions. A multi-tiered server
structure combines these server types into clusters for seamless client access.
The major server tiers include
➤ Front-end servers

➤ Mid-tier servers

➤ Back-end servers

Understand the name and function of each tier of a multi-tiered server.

01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 26

26 Chapter
. . . . .1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Front-End Servers
Front-end servers function similarly to appliance servers. They can be config-
ured to perform one function, or a multiple of related functions, and are
treated as field-replaceable units. Typical front-end servers are used for
➤ Mail servers

➤ Proxy servers

➤ Firewall servers

➤ Web services

Front-end servers are specifically configured to pull information from the

organization’s mid-tier and back-end servers, such as data being presented by
the front-end server’s web services. To quickly retrieve and display the data
from the other tiers on the web page, the front-end server is configured with
a large amount of installed RAM to accommodate the necessary volume of
data efficiently.
Front-end servers also provide and control the external or remote access to
the company network. This is normally accomplished through the telephone
system using dial-up connections. Windows server OSs manage this function
through a Remote Access Service (RAS) utility, the client portion of which is
sometimes called Dial-up Networking.
Before adding a dial-up connection to a network, additional hardware and
configuration effort must be applied to the server. An even greater concern
in adding dial-up access to the network is the security exposure from open-
ing the network to access by outsiders.
The network administrator must balance the advantages of granting access
to trusted remote users—such as outside sales people, traveling personnel,
and work-at-home personnel—against the risks posed by intruders and
hackers. Additional security-related hardware and server configuration set-
tings are required to compensate for these possibilities.

Be aware of factors that should be considered before adding a dial-up connection to

a network.

Remote access can also be provided through an Internet connection using a

Virtual Private Network (VPN). Secure VPN communications are provided
through a tunneling protocol that encrypts the data, so that only the in-
tended receiver can decrypt and read it.
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 27

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.0—Server
. . . . . . .Basics
27
. . .

Mid-Tier Servers
Mid-tier servers are used to process and relay information between front-end
and back-end servers. Using hardware very similar to that described for
general-purpose servers, most mid-tier servers are designed to perform more
than one function. They may act as both a file server and as a mail server, or
designed to act as a back-end or a front-end server. A mid-tier server is nor-
mally too powerful to be used simply as a front-end server, and not powerful
enough to perform as a reliable back-end server.
In larger companies, the mid-tier is usually made up of network servers. In
very large organizations, the middle tier may include database and web
servers that pull data from back-end servers Fault tolerance is provided by
including redundant components inside the chassis, and using redundant
machines throughout the tier.
Mid-tier machines are more expensive than front-end servers, and the pur-
chase of offline backup machines is not cost-effective. Instead, larger com-
panies often install a second machine to perform the identical function. If
one mid-tier machine fails, the redundancy from the online backup guards
against losing the resource. By adding the second machine online, the over-
all performance of the middle tier for that function is increased.

Back-End Servers
Back-end servers typically are large, expensive units used to store vast volumes
of data in archive and data farms. A typical back-end server used in a data
warehouse operation may have 30 or more disk drives connected to it.
Because they hold the data required by working mid-tier and front-end
servers, back-end servers are usually required to be up and running 99.99%
of the time.

A data warehouse is a computer storage system containing a wide variety of data.

Information is combined from many different databases across an entire organization.
These storage structures are used to track business conditions at any single point in
time. Systems for extracting data from large databases provide IT managers with flex-
ible methods of access. Keep in mind that a data farm is synonymous with a data
warehouse.

An archive server holds all of the data that has previously been backed up and
archived. This differs from a traditional backup where data is transferred to
some type and then stored in a secure area. Retrieving traditional data back-
ups requires physically going to the secure area and inserting the media into
the backup server. An archive server will store all the data to some type of
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 28

28 Chapter
. . . . .1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

media, and will allow the archived data to be accessed without manual inter-
vention. For tape backups, the server would access the specific tape contain-
ing the required data using a mechanical arm to retrieve and insert the tape
into the server.
Back-end machines are so expensive that procuring an additional standby
machine is not a cost–effective solution. Instead, a high level of redundancy
is built into the chassis of a back-end server, including power-on fault toler-
ance, in order to keep the system up and running even when replacing failed
components.

Large network systems often provide multiple levels of data backup. The first, and
most efficient, backup level is directly on a local hard disk drive. The second level of
archival storage is reserved for data that is not used immediately. The first level of
backup data may periodically be archived on a secondary storage device, such as a
tape drive. In large networks, the archival storage function may be performed and
controlled by a server, or a cluster of servers, referred to as archival servers.

Back-end servers do not require large amounts of RAM, but they do require
fast disk access, plenty of processing power, and high storage capacities. In
data farms, the servers may be organized into Storage Area Networks
(SANs). SANs are specialized, high-speed fiber optic networks that extract
data from the servers without using the main network’s bandwidth. They are
connected directly to the back-end servers, and pull the data up to the main
network.

Implementing Server Tiers

How servers are implemented in an organization is usually determined by the
size of the organization and what resources are required. Smaller companies
may not implement a fully tiered front-end, mid-tier, and back-end server
network. They are more likely to build their network structure around a clus-
ter of general-purpose servers, where each server performs many tasks. One
server might be designated as a proxy server providing all network users with
Internet access. It may also perform as the mail server, and also take on the
intranet server chores. The company may also have another general-purpose
server, configured and designated to provide database and file services.
Large organizations design their networks with all three tiers of the multi-
tier server model. The construction and function of each server tier are based
on the applications they are to provide in the network structure. The front-
end machines are used to provide the interface between the network and the
users, such as customers and employees. This tier is typically made up of
multiple servers and redundant machines, to ensure that the required re-
sources are always available.
01_0789733684_CH01.qxd 10/18/05 11:31 AM Page 29

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.0—Server
. . . . . . .Basics
29
. . .

As indicated earlier, the middle tier is made up of servers that act as inter-
mediates between the front-end and back-end tiers. Like the front-end tier,
the middle tier normally consists of multiple machines and redundant
machines. However, the servers in this tier also tend to include a high level
of redundancy inside the chassis itself.
The back-end tier is made up of large, expensive machines that typically rely
on redundancy inside the system to keep them available. Their high cost
tends to make using separate redundant units unrealistic. As you can see, the
reasoning behind the types of redundancy employed at each level is driven
primarily by cost.
Some mid-sized companies may skip either the back-end or front-end server
tiers depending on their current needs. Other medium-sized companies may
configure a portion of their multi-tiered servers to perform at more than one
level. For example, a company might use what would be considered mid-
tiered machines to perform front-end and mid-tier tasks, while their back-
end servers simply perform back-end functions. This arrangement, as de-
picted in Figure 1.3, would save the company money without totally com-
promising their resources. In addition, the structure could later be expanded
into a fully tiered network as the company’s needs grow.

Front-end
Tasks: Mid-tier
Tasks:
Proxy server
Mail server File server
Mail server Back-end server
Firewalls (data server)
Web services Database server

Front-end/
mid-tier
server

Client Client Client

Figure 1.3 A modified tier server structure.