Chapter 10

Setting Up a Server

In This Chapter

Getting a handle on network operating system features

Discovering Windows Server and other server options

Thinking about the different ways to install a network operating system

Getting ready for the installation

Installing a network operating system

Configuring your server roles

One of the basic choices that you must make before you proceed any further in building your network is to decide which network operating system (NOS) to use as the foundation for your network. This chapter begins with a description of several important features found in all network operating systems. Next, it provides an overview of the advantages and disadvantages of the most popular NOSes.

Of course, your work doesn’t end with the selection of an NOS. You must then install and configure the operating system to get it working. This chapter provides an overview of what’s involved with installing and configuring Microsoft’s latest and greatest server operating system, Windows Server 2012.

Network Operating System Features

All NOSes must provide certain core functions, such as connecting to other computers on the network, sharing files and other resources, and providing for security. In the following sections, I describe some core NOS features in general terms.

Network support

It goes without saying that an NOS should support networks. That requires a range of technical capabilities:

An NOS must support a wide variety of networking protocols to meet the needs of its users.

A large network typically consists of a mixture of various versions of Windows as well as Macintosh and Linux computers. As a result, the server may need to simultaneously support TCP/IP, NetBIOS, and AppleTalk protocols.

Many servers have more than one network interface card (NIC) installed. In that case, the NOS must be able to support multiple network connections:

• Ideally, the NOS should be able to balance the network load among its network interfaces.

• If one of the connections fails, the NOS should be able to seamlessly switch to another connection.

Most NOSes include a built-in capability to function as a router that connects two networks.

The NOS router functions should also include firewall features to keep unauthorized packets from entering the local network.

File-sharing services

One of the most important functions of an NOS is to share resources with other network users. The most common resource that’s shared is the server’s file system — organized disk space that a network server must be able to share (in whole or in part) with other users. In effect, those users can treat the server’s disk space as an extension of their own computers’ disk space.

The NOS allows the system administrator to determine which portions of the server’s file system to share.

Although an entire hard drive can be shared, it isn’t commonly done. Instead, individual folders are shared. The administrator can control which users are allowed to access each shared folder.

Because file sharing is the reason why many network servers exist, NOSes have more sophisticated disk management features than are found in desktop OSes. For example, most NOSes can manage two or more hard drives as though they were a single drive. In addition, most can create a mirror — an automatic backup copy of a drive — on a second drive.

Multitasking

Only one user at a time uses a desktop computer; however, multiple users simultaneously use server computers. As a result, an NOS must provide support for multiple users who access the server remotely via the network.

At the heart of multiuser support is multitasking — a technique that slices processing time microthin and juggles the pieces lightning fast among running programs. It’s how an OS can execute more than one program (a task or a process) at a time. Multitasking operating systems are like the guy who used to spin plates balanced on sticks on the old Ed Sullivan Show: running from plate to plate, trying to keep them all spinning so that they don’t fall off the sticks. To make it challenging, he’d do it blindfolded or riding on a unicycle. Substitute programs for the plates and file management for the unicycle, and there you are.

Although multitasking creates the appearance that two or more programs execute on the computer at the same time, in reality, a computer with a single processor can execute only one program at a time. The OS switches the CPU from one program to another to create the appearance that several programs execute simultaneously, but at any given moment, only one program processes commands. The others are patiently waiting their turns. (However, if the computer has more than one CPU core, the CPU cores can execute programs simultaneously — but that’s another kettle of fish.)

Directory services

Directories are everywhere — and were, even in the days when they were all hard copy. When you needed to make a phone call, you looked up the number in a phone directory. When you needed to find the address of a client, you looked her up in your Rolodex. And then there were the nonbook versions: When you needed to find the Sam Goody store at a shopping mall (for example), you looked for the mall directory — usually, a lighted sign showing what was where.

Networks have directories, too, providing information about the resources that are available on the network: users, computers, printers, shared folders, and files. Directories are essential parts of any NOS.

In early NOSes (such as Windows NT 3.1 and NetWare 3.x), each server computer maintained its own directory database — a file that contained an organized list of the resources available just on that server. The problem with that approach was that network administrators had to maintain each directory database separately. That wasn’t too bad for networks with just a few servers, but maintaining the directory on a network with dozens or even hundreds of servers was next to impossible.

In addition, early directory services (programs that made the directory databases usable) were application-specific. For example, a server had one directory database for user logons, another for file sharing, and yet another for e-mail addresses. Each directory had its own tools for adding, updating, and deleting directory entries.

The most popular modern directory service is Active Directory (AD), which is standard with Windows-based server OSes. Active Directory provides a single directory of all network resources. It drops the old-style 15-character domain and computer names used by Windows NT Server in favor of Internet-style DNS-style names, such as Marketing.MyCompany.com or Sales.YourCompany.com. Figure 10-1 shows the Active Directory Users and Computers tool, which manages Active Directory user and computer accounts on Windows Server 2008.

Figure 10-1: Active Directory Users and Computers.

Security services

All NOSes must provide some measure of security to protect the network from unauthorized access. Hacking seems to be the national pastime these days. With most computer networks connected to the Internet, anyone anywhere in the world can — and probably will — try to break into your network.

The most basic type of security is handled through user accounts, which grant individual users the right to access the network resources and govern which resources the user can access. User accounts are secured by passwords; therefore, good password policy is a cornerstone of any security system. Most NOSes give you some standard tools for maintaining network security:

Establish password policies. For example, you can mandate that passwords have a minimum length and include a mix of letters and numerals.

Set passwords to expire after a certain number of days. Network users must change their passwords frequently.

Encrypt network data. A data-encryption capability scrambles data before it’s sent over the network or saved on disk, making unauthorized use a lot more difficult.

Good encryption is the key to setting up a virtual private network (VPN), which enables network users to securely access a network from a remote location by using an Internet connection.

Issue digital certificates. These special codes are used to ensure that users are who they say they are and files are what they claim to be.

Microsoft’s Server Operating Systems

Over the years, Microsoft has released several versions of its Windows-based server OS: Windows NT Server 4, Windows 2000 Server, Windows 2003 Server, and Windows Server 2008 — and now Windows Server 2012. Most organizations are still using Windows Server 2008 or 2003. In fact, plenty of organizations still use Windows 2000 Server, and a few (mostly on deserted islands cut off from civilization) run Windows NT Server 4.

---

---

Each new version builds on the previous version by introducing new and improved features. However, keep in mind as you read the following sections that Windows NT Server 4, Windows 2000 Server, and Windows 2003 Server are considered obsolete.

Windows NT Server 4

Windows NT Server was the last in a long series of Windows servers dubbed NT, which stood for New Technology. The “new technology” that got everyone so excited about Windows NT in the first place was 32-bit processing, which was a huge step from the 16-bit processing of earlier versions of Windows. Windows NT was the first Microsoft OS that was reliable enough to work as a network server on large networks. At more than 16 years old, Windows NT 4 is hopelessly obsolete, but it’s still important because of its legacy. Many OS features that were first introduced as part of Windows NT 4 are still in use today.

Probably the most important feature of Windows NT is its directory model, which is based on the concept of domains, which are groups of computers managed by a single directory database. To access shared resources within a domain, you must have a valid user account within the domain and be granted rights to access the resources in which you’re interested. The domain system uses 15-character NetBIOS names to access individual computers within a domain and to name the domain itself.

Windows 2000 Server

Windows 2000 Server, built on the strengths of Windows NT Server 4, was faster, easier to manage, more reliable, and easier to use for large and small networks alike.

The most significant new feature offered by Windows 2000 Server was Active Directory (AD), which provides a single directory of all network resources and enables program developers to incorporate the directory into their programs. Active Directory drops the 15-character domain and computer names in favor of Internet-style DNS names, such as Marketing.MyCompany.com or Sales.YourCompany.com. (However, it still supports the old-style names for older clients that don't deal well with DNS names.)

Windows 2000 Server came in three versions:

Windows 2000 Server was the basic server, designed for small to medium-sized networks. It included all the basic server features, including file and printer sharing, and acted as a web and e-mail server.

Windows 2000 Advanced Server was the next step up, designed for larger networks. Advanced Server could support server computers that had up to 8GB of memory (not hard drive — RAM!) and four integrated processors instead of the single processor that desktop computers and most server computers had.

Windows 2000 Datacenter Server supported servers that had as many as 32 processors with up to 64GB of RAM and was specially designed for large database applications.

For small networks with 50 or fewer computers, Microsoft offered a special Small Business Server bundle, which included the following components for one low, low price:

Windows Server 2003: The OS for your network server

Exchange Server 2003: For e-mail and instant messaging

SQL Server 2000: A database server

FrontPage 2000: For building websites

Outlook 2000: For reading e-mail

Windows 2003 Server

The next server version of Windows was Windows 2003 Server, which was built on Windows 2000 Server, with the following added features:

A new-and-improved version of AD with tighter security, an easier-to-use interface, and better performance.

A better and easier-to-use system management interface: the Manage My Server window. On the flip side, for those who prefer brute-force commands, Windows 2003 Server includes a more comprehensive set of command line management tools than is offered by Windows 2000 Server. Of course, the familiar Microsoft Management Console tools from Windows 2000 Server are still there.

A major change in the application-programming interface for Windows programs, known as the .NET Framework.

Support for ever-larger clusters of computers. A cluster is a set of computers that work together as if they were a single server. Windows 2000 Server Datacenter Edition and previous versions supported clusters of four servers; Windows 2003 Server Enterprise and Datacenter Editions support clusters of eight servers. (Obviously, this is a benefit only for very large networks. The rest of us should just grin and say, “Cool!”)

An enhanced distributed file system that lets you combine drives on several servers to create one shared volume.

Support for storage area networks (SANs).

A built-in Internet firewall to secure your Internet connection.

A new version of Microsoft’s web server, Internet Information Services (IIS) 6.0.

Like its predecessor, Windows 2003 Server shipped in several versions:

Windows 2003 Server, Standard Edition: This is the basic version of Windows 2003. If you’re using Windows Server 2003 as a file server or to provide other basic network services, this is the version you’ll use. Standard Edition can support servers with up to four processors and 4GB of RAM.

Windows 2003 Server, Web Edition: A version of Windows 2003 optimized for use as a web server.

Windows 2003 Server, Enterprise Edition: Designed for larger networks, this version can support servers with up to eight processors, 32GB of RAM, server clusters, and advanced features designed for high performance and reliability.

Windows 2003 Server, Datacenter Edition: The most powerful version of Windows 2003, with support for servers with 64 processors, 64GB of RAM, and server clusters, as well as advanced fault-tolerance features designed to keep the server running for mission-critical applications.

Windows Server 2008

In February of 2008, Microsoft released the successor to Windows 2003 Server, not surprisingly known as Windows Server 2008. Windows Server 2008 added many new features, including the following:

Even more enhancements to AD, including the ability to manage digital certificates, a new type of domain controller (a read-only domain controller), and the ability to stop and restart AD services without shutting down the entire server.

A new graphical user interface (GUI) based on Windows Vista, including a new all-in-one management tool: Server Manager.

A new version of the OS called Server Core, which has no GUI. Server Core is run entirely from the command line or by a remote computer that connects to the server via Microsoft Management Console. Server Core is designed to provide efficient file servers, domain controllers, or DNS and DHCP servers.

Remote connection enhancements that enable computers to establish web-based connections to the server using the HTTPS protocol without having to establish a virtual private network (VPN) connection.

Yet another new version of the Internet Information Services (IIS) web server (7.0).

Windows Server 2008 R2

In the fall of 2009, Microsoft issued an update to Windows Server 2008, officially called Windows Server 2008 R2. Network administrators the world over rejoiced, in part because most of them are also Star Wars fans and they can now refer to their favorite operating system as “R2.”

R2 builds on Windows Server 2008 with a variety of new features, including virtualization features that let you run more than one instance of the OS on a single server computer, a new version of IIS (7.5), and support for up to 256 processors.

Also, R2 officially drops support for 32-bit processors. In other words, R2 runs only on server-class 64-bit processors, such as Intel Itanium and Xeon.

Windows Server 2012

The newest version of Windows Server, officially known as Windows Server 2012, offers many significant improvements over Windows Server 2008, the most notable being the new Metro user interface (UI), which is designed for use with touch-sensitive displays.

Other new features of Windows Server 2012 include

A new file system – ReFS – that replaces NTFS, providing better performance and reliability

A redesigned Task Manager designed to highlight which system tasks are drawing more of the server’s CPU, memory, and disk and network I/O capacity

IP Address Management, which is a feature designed to automatically discover what IP addresses are being used by computers and other devices on the network

The ability to support servers with as many as 640 processors and 4TB of RAM

Seeing Other Server Operating Systems

Although Windows Server is the most popular choice of NOS, you have other choices. The following sections briefly describe two other server choices: Linux and Macintosh OS X Server.

Linux

Perhaps the most interesting OS now available is Linux. The free Linux OS is based on Unix, which is a powerful NOS often used on large networks. Linux was started by Linus Torvalds, who thought it’d be fun to write a version of Unix in his free time — as a hobby. He enlisted help from hundreds of programmers throughout the world, who volunteered their time and efforts via the Internet. Today, Linux is a full-featured version of Unix; its users consider it to be as good as or better than Windows.

Linux offers the same networking benefits of Unix and can be an excellent choice as a server OS.

Mac OS X Server

All the other server OSes I describe in this chapter run on Intel-based PCs with Pentium or Pentium-compatible processors. But what about Macintosh computers? After all, Macintosh users need networks, too. For Macintosh networks, Apple offers a special NOS — Mac OS X Server — which has all the features you expect, including file and printer sharing, Internet features, e-mail, and more.

Novell NetWare

NetWare was once the king of network operating systems. Today, NetWare networks are rare, but you can still find them if you look hard enough. NetWare has always had an excellent reputation for reliability. In fact, some network administrators swear that they have NetWare servers on their networks that have been running continuously, without a single reboot, since Ronald Reagan was president. (Unfortunately, there hasn’t been a major upgrade to NetWare since George W. Bush’s first term.)

Novell released the first version of NetWare in 1983, two years before the first version of Windows and four years before Microsoft’s first network operating system, the now defunct LAN Manager. Over the years, NetWare has gone through many versions. The most important versions were

NetWare version 3.x, the version that made NetWare famous. NetWare 3.x used a now-outdated directory scheme called the bindery. Each NetWare 3.x server has a bindery file that contains information about the resources on that particular server. With the bindery, you had to log on separately to each server that contained resources you wanted to use.

NetWare 4.x, in which NetWare Directory Service (NDS) replaced the bindery. NDS is similar to AD. It provides a single directory for the entire network rather than separate directories for each server.

NetWare 5.x was the next step, with a new UI based on Java for easier administration, improved support for Internet protocols, multiprocessing with up to 32 processors, and many other features.

NetWare 6.0 introduced a variety of new features, including a new disk management system (Novell Storage Services); web-based access to network folders and printers; and built-in support for Windows, Linux, Unix, and Macintosh file systems.

Novell released its last major version of NetWare (6.5) in summer 2003. It included improvements to its browser-based management tools and was bundled with open source servers, such as Apache and MySQL.

Beginning in 2005, NetWare has transformed itself into a Linux-based system: Open Enterprise System (OES). In OES, the core of the OS is actually Linux, with added applications that run the traditional NetWare services such as directory services. (For more information, see “Linux,” earlier in this chapter.)

The Many Ways to Install a Network Operating System

Regardless of which NOS you choose to use for your network servers, you can use any of several common ways to actually install the NOS software on the server computer. The following sections describe these alternatives.

Full install versus upgrade

One of the basic NOS installation choices is whether you want to perform a full installation or an upgrade installation. In some cases, you may be better off performing a full installation even if you’re installing the NOS on a computer with an earlier version of the NOS installed:

If you’re installing the NOS on a new server, you’ll be performing a full installation that installs the OS and configures it with default settings.

If you’re installing the NOS on a server computer that already has a server OS installed, you can perform an upgrade installation to replace the existing OS with the new one, yet retain as many of the settings from the existing OS as possible.

You can also perform a full installation on a computer that already has an OS installed. In that case, you have the option of deleting the existing OS or performing a multiboot installation that installs the new server OS alongside the existing OS. Then, when you restart the computer, you can choose which OS you want to run.

  Although multiboot installation may sound like a good idea, it’s fraught with peril. Avoid multiboot unless you have a specific reason to use it.

You can’t upgrade a client version of Windows to a server version. Instead, you must perform a full installation, which deletes the existing Windows OS, or a multiboot installation, which leaves the existing client Windows intact. Either way, however, you can preserve existing data on the Windows computer when you install the server version.

Installing over the network

Typically, you install the NOS directly from the CD-ROM distribution discs on the server’s CD-ROM drive. However, you can also install the OS from a shared drive located on another computer — provided that the server computer already has access to the network. You can either use a shared CD-ROM drive or copy the entire contents of the distribution CD-ROM disc onto a shared hard drive.

If you’re going to install the NOS onto more than one server, save time by first copying the distribution CD onto a shared hard drive. That’s because even the fastest CD-ROM drives are slower than the network. Even with a slow 100 Mbps network, access to hard drive data over the network is much faster than access to a local CD-ROM drive.

Gathering Your Stuff

Before you install an NOS, gather everything you need so you don’t have to look for something in the middle of the setup. The following sections describe the items you’re most likely to need.

A capable server computer

Obviously, you have to have a server computer on which to install the NOS. Each NOS has a list of minimum hardware requirements supported by the OS. For example, Table 10-1 summarizes the minimum requirements for Windows Server 2012.

My suggestion is that you take these minimums with a grain of salt. Windows Server 2012 will crawl like a snail with 512MB of RAM; I wouldn’t bother with less than 4GB, and 16GB is a more appropriate minimum for most purposes.

Table 10-1 Minimum Hardware Requirements for Windows Server 2012

Item	Windows Server 2012
CPU	1.4 GHz
RAM	512MB
Free disk space	32GB

Also check your server hardware against the list of compatible hardware published by the maker of your NOS. For example, Microsoft publishes a list of hardware that it has tested and certified as compatible with Windows servers. You can find this list at www.windowsservercatalog.com.

The server OS

You also need a server OS to install. You need the distribution CDs or DVDs or access to copies of them over the network. In addition to the discs, you should have the following:

The product key: The installation program will ask you to prove that you have a legal copy of the software. If you have the actual CDs or DVDs, the product key should be on a sticker attached to the case.

Manuals: If the OS came with printed manuals, you should keep them handy. If the manuals are in PDF form, keep the PDFs handy.

Your license type: You can purchase Microsoft OSes on a per-server or a per-user/per-device basis. You need to know which plan you have when you install the NOS.

Check the CD or DVD distribution disc for product documentation and additional last-minute information. For example, Windows servers have a docs folder that contains several files that have useful setup information.

Other software

In most cases, the installation program should be able to automatically configure your server’s hardware devices and install appropriate drivers. Just in case, though, you should dig out the driver disks/discs that came with your devices, such as network interface cards (NICs), SCSI devices, DVD drives, printers, scanners, and so on.

A working Internet connection

Online connectivity isn’t an absolute requirement, but the installation will go much smoother if you have a working Internet connection before you start. The installation process may use this Internet connection for several things:

Downloading late-breaking updates or fixes to the OS: This can eliminate the need to install a Service Pack after you finish installing the NOS.

Locating drivers for nonstandard devices: This can be a big plus if you can’t find the driver disk for an obscure SCSI card.

Activating the product after you complete the installation (for Microsoft OSes)

A good book

You’ll spend lots of time watching progress bars during installation, so you may as well have something to do while you wait. I recommend Les Miserables.

Making Informed Decisions

When you install a NOS, you have to make some decisions about how you want the OS and its servers configured. Most of these decisions aren’t cast in stone, so don’t worry if you’re not 100 percent sure how you want everything configured. You can always go back and reconfigure things. However, you’ll save yourself time if you make the right decisions up front rather than just guess when the setup program starts asking you questions.

The following list details most of the decisions that you’ll need to make. (This list is for Windows Server 2012 installations. For other network OSes, the decisions may vary slightly.)

The existing OS: If you want to retain the existing OS, the installation program can perform a multiboot setup, which allows you to choose which OS to boot to each time you start the computer. This is rarely a good idea for server computers, so I recommend that you elect to delete the existing OS.

Partition structure: Most of the time, you’ll want to treat the entire server disk as a single partition. However, if you want to divide the disk into two or more partitions, you should do so during setup. (Unlike most of the other setup decisions, this one is hard to change later.)

Computer name: During the OS setup, you’ll be asked to provide the computer name used to identify the server on the network. If your network has only a few servers, you can just pick a name such as Server01 or MyServer. If your network has more than a few servers, you’ll want to establish a naming convention you can follow for naming your servers.

Administrator password: Okay, this one is tough. You don’t want to pick something obvious, like Password, Administrator, or your last name. On the other hand, you don’t want to type in something random that you’ll later forget because you’ll find yourself in a big pickle if you forget the administrator password. I suggest that you make up a complex password with uppercase and lowercase letters, some numerals, and a special symbol or two; then write it down and keep it in a secure location where you know it won’t get lost.

Networking protocols: You’ll almost always need to install the TCP/IP protocol, the Microsoft network client protocol, and file and printer sharing. Depending on how the server will be used, you may want to install other protocols as well.

TCP/IP configuration: You’ll need to know what IP address to use for the server. Even if your network has a DHCP server to dynamically assign IP addresses to clients, most servers use static IP addresses.

Workgroup or domain: You’ll need to decide whether the server will join a domain or just be a member of a workgroup. In either case, you need to know the domain name or the workgroup name. (In most cases, if you are installing Windows Server, you’ll be using a domain. Workgroups are mostly used for peer-to-peer networks that do not have dedicated servers.)

Final Preparations

Before you begin the actual installation, take a few more steps:

Tidy up. Clean up the server’s disk by uninstalling any software that you don’t need and removing any old data that is no longer needed. This cleanup is especially important if you’re converting a computer that’s been in use as a client computer to a server. You probably don’t need Microsoft Office or a bunch of games on the computer after it becomes a server.

Backup. Do a complete backup of the computer. Operating system setup programs are almost flawless, so the chances of losing data during installation are minimal, but you still face the chance that something may go wrong.

Disconnect serial and USB connection. If the computer is connected to an uninterruptible power supply (UPS) that has a serial or USB connection to the computer, unplug the serial or USB connection. In some cases, this control connection can confuse the OS’s setup program when it tries to determine which devices are attached to the computer.

Uncompress drives. If the computer has hard drives compressed with DriveSpace or DoubleSpace, uncompress the drives before you begin.

Chill out. Light some votive candles, take two acetaminophen, and put on a pot of coffee.

Installing a Network Operating System

The following sections present an overview of a typical installation of Windows Server 2012. Although the details vary, the overall installation process for other NOSes is similar.

In most cases, the best way to install Windows Server 2012 is to perform a new install directly from the DVD installation media. Although upgrade installs are possible, your server will be more stable if you perform a new install. (For this reason, most network administrators avoid upgrading to Windows Server 2012 until it’s time to replace the server hardware.)

To begin the installation, insert the DVD distribution media in the server’s DVD drive and then restart the server. This causes the server to boot directly from the distribution media, which initiates the setup program.

As the setup program proceeds, it leads you through two distinct installation phases: Collecting Information and Installing Windows. The following sections describe these installation phases in greater detail.

Phase 1: Collecting Information

In the first installation phase, the setup program asks for the preliminary information that it needs to begin the installation. A setup wizard prompts you for the following information:

Language: Select your language, time zone, and keyboard type.

Product Key: Enter the 25-character product key that came with the installation media. If setup says that you entered an invalid product key, double-check it carefully. You probably just made a typo.

Operating System Type: Select Windows Server 2012 Standard Edition or Core.

• Standard Edition: Install the full server OS.

• Core: Install the new text-only version. Use this version only if you are a master of the Windows Server command-line interface known as PowerShell.

License Agreement: The official license agreement is displayed. You have to agree to its terms to proceed.

Install Type: Choose an Upgrade or Clean Install type.

Disk Location: Choose the partition in which you want to install Windows.

Phase 2: Installing Windows

In this phase, Windows setup begins the actual process of installing Windows. The following steps are performed in sequence:

1. Copying Files: Compressed versions of the installation files are copied to the server computer.

2. Expanding Files: The compressed installation files are expanded.

3. Installing Features: Windows server features are installed.

4. Installing Updates: The setup program checks the Microsoft website and downloads any critical updates to the OS.

5. Completing Installation: When the updates are installed, the setup program reboots so it can complete the installation.

Configuring Your Server

After you install Windows Server 2012, the computer automatically reboots, and you’re presented with the Server Manager program as shown in Figure 10-2.

Figure 10-2: The Windows 2012 Server Manager.

From Server Manager, you can perform a number of tasks that are necessary to configure the server for production use. Specifically, you can configure the server roles: the networking features that the server will provide, such as file server, web server, DHCP server, and DNS server.