How you choose to install Windows depends on what is already on the server, where the installation files are, and how many installations you have to do.
If you only need to perform a few installations, you can manually install Windows, either from within an existing copy of Windows or by booting from the Windows CD-ROM.
If you need to perform multiple installations, you can create unattended answer files to automate the setup process, use SysPrep to clone a computer with Windows and applications already installed, or use Remote Installation Services (RIS) to perform automated installations from across the network (RIS is discussed in Chapter 25).
The most basic way of installing Windows is to manually go through each Setup screen (instead of automating the process). This is fine when doing a few installations or when you're familiarizing yourself with the installation process. However, it's a tedious and slow method of deploying systems en masse, so for multiple installations, automate the process using answer files, SysPrep, or RIS.
The Phases of Setup
The Windows setup process consists of several phases that vary somewhat depending on how you initiate the installation.
If you install from Windows, Setup gathers information and copies the files that are necessary for the computer to boot into text-mode setup, which it then reboots into. You then (optionally) select the appropriate partition, after which Setup installs a minimal version of Windows to the hard disk and reboots into the graphical user interface (GUI)-based Windows Setup Wizard, which collects more information, configures the devices, and finishes copying files. After this, Setup is complete, and the computer reboots into Windows proper.
If you boot the system from the Windows CD-ROM, Setup launches straight into text-mode setup, where you select the partition on which you want to install Windows. Setup then installs Windows to the hard disk and then boots into the GUI-based Windows Setup Wizard. The Setup Wizard then collects information, configures devices, and finishes installing files to the computer. Setup reboots the computer when it's finished.
If you boot the system with a Windows 98 boot disk and then run Setup, Setup copies the necessary files to the hard disk to boot into text-mode setup and then reboots into text-mode setup. Setup then performs the same process it would perform if you booted from a Windows CD-ROM or (or setup boot disks).
If you already have a copy of Windows 2000, Windows NT 4, or Windows 98 installed on the server, the easiest way to install Windows is to run the 32-bit version of setup (Winnt32.exe). To do so, follow these steps (if there isn't already a copy of Windows on the computer, see the sidebar "Installing Windows on a Blank Disk"):
Figure 5-1. The Windows AutoPlay window.
Figure 5-2. Windows Setup.
Figure 5-3. The Select Special Options window.
Figure 5-4. The Advanced Options dialog box.
You can change the language options after installation by using Control Panel.
Real World
Installing Windows on a Blank Disk
If you don't have a copy of Windows already installed on the server, you need to boot the system with the Windows CD-ROM. To boot from the Windows CD-ROM, insert the CD into the drive and reboot the system. When prompted, press the any key (just teasing—press a key) to boot from the CD. If Setup doesn't launch automatically, you might need to configure the BIOS boot order to tell the system to use the CD-ROM drive before the hard disk drive.
If you are unable to boot from the CD-ROM or you need to install from a network share, you can use the Windows 2000 Setup boot disks, or you can use a boot disk created with Windows 98—the only solution if you want to install from a network source (although Windows XP and the Windows .NET Server family can create bootable floppy disks, these disks don't contain SmartDrive, making the installation process hours longer than it should be). After booting from the floppy, launch the Winnt.exe program from the \i386 folder on the Windows CD-ROM or network source (the floppy disk needs to have appropriate network drivers loaded to access network shares).
After booting from the Setup boot disks or launching the Winnt.exe program the computer reboots into the text-based phase of Windows Setup, as described in the next section (make sure to eject the floppy disk before the reboot).
Figure 5-5. Choosing a disk partition for Windows 2000.
If you have a removable drive with a capacity of 829 MB or more, you can install Windows onto a removable disk. This step is recommended only for parallel installations for recovery or testing purposes, not for the primary installation of Windows.
Deleting a partition erases all information on that partition. Don't delete a partition unless all data on the partition is reliably backed up.
The hardware detection part of the setup process is where problems occur most frequently. If the system becomes unresponsive or doesn't progress for a long time (give it 30 minutes to an hour if possible), press the Reset button on the computer to restart the computer. Setup automatically resumes where it left off, and usually completes Setup successfully.
You can use the Regional Options tool in Control Panel to change regional settings after you install Windows, so you probably don't need to linger over these dialog boxes.
Figure 5-6. Per Server and Per Seat licensing modes.
You can use the Licensing Control Panel tool after installing Windows to modify the licensing options.
Press Shift+F10 at any time during the Windows Setup Wizard to open a command prompt from which you can do devious things, such as inspect log files or launch a screen capture program.
Real World
Password Security
To make the system as secure as possible, always assign a password to the administrator account, preferably a password at least seven characters long and consisting of mixed letters and special characters, uppercase and lowercase. We like to use acronyms for phrases that are meaningful to us, easy to remember, and unlikely to be meaningful or memorable to anyone else, such as Uk,Ur?Ue! (You know, you are what you eat!)
You should also clear the logon history after installing Windows so that would-be hackers must figure out both the password and the user name. Another good precaution after installation is to use the built-in Administrator account to make a second account with full administrative privileges. This account can have a generic name, or you can call it something descriptive. Use it for all the day-to-day administrative work. Assign a special secure password to the built-in Administrator account and rename it from the default. Stash the password and name somewhere safe and relegate that account to semiretirement. Because it's possible to disable any administrator account, including the built-in Administrator account, it's wise to have a backup account. That way, you'll always have an uncontaminated known-to-be-good administrator account to which you can resort, just in case.
Internet Information Services (IIS) is installed by default and opens the server to viruses and security exploitation (although this isn't as big a deal on a private network, there's still no point in opening up more security risks). We recommend that you not install IIS during Windows setup, even if the server is destined to become a Web server, unless the Windows installation files already have the latest service pack applied.
Figure 5-7. The Windows 2000 Components window.
Figure 5-8. The Networking Components window.
Don't disable NetBIOS over TCP/IP, as doing so disables browsing and causes problems logging onto Windows NT 4 domains. If Windows Internet Name Service (WINS) is used on the network, make sure to enable it if TCP/IP settings aren't obtained from a DHCP server.
Figure 5-9. The Domain Membership window.
Real World
Naming Computers
Take a few moments to think about naming conventions before you commit yourself. Many ways exist to create names on the network, from the cute to the whimsical to the sensible. Sometimes system administrators devise arbitrary schemes based on algorithms known only to them. Or they attempt to insert charm into the process of computer naming. Block those impulses!
It's easy for you to keep a map of what and where the different clients and servers are on the network, but if you make life hard on users, you pay in the long run. So naming all the domain controllers after Shakespearean characters and naming the client machines after characters in Greek mythology might make sense to you. And it does, of course, have a certain poetic grandeur. But it isn't going to help users figure out that "Cordelia" is the server in Legal and "Goneril" is the one in Production. On the other hand, using "legal_srv" for the server in the Legal department tells everyone immediately which machine it is.
Try to resist the desire to be cute with your machine names. And at all costs, avoid arbitrary names based on some formula that only you know how to crack.
Installing Windows 2000, like installing any operating system, takes a lot of time. Setting up multiple servers can wipe out entire days if you set up each server manually. The time you can save by automating installations for a large number of similar servers far outweighs the occasional hiccups you might encounter.
Before automating installations, run through a couple of manual installations to get a better feel for the installation process. Review the earlier parts of this chapter and pay attention during the installations. Then try using the command-line switches to partially automate Setup. Following that, use the Setup Manager Wizard to create and test the fully or partially automated setup scripts.
Don't feel that you must fully automate all setups. Using automated answer files (setup scripts) can provide significant time savings, but it's important to choose the appropriate level of automation for the particular job. That might involve creating a single answer file that you'll use to deploy hundreds of servers, or it might mean using just one or two command-line parameters to speed up the installation of a single server.
You can streamline the setup process on a single machine by launching Setup using command-line parameters; you also use command-line parameters to specify an answer file to completely automate Setup.
To use a command-line parameter on a computer with Windows, boot the computer into Windows and open a command prompt window. Then type [path]\winnt32.exe[parameter], substituting [path] with the location of the Windows setup files and replacing [parameter] with the appropriate parameter or parameters you want to use. Table 5-3 shows the available parameters and what they do. (These are for use with only Winnt32.exe, the 32-bit version of Setup.)
Table 5-3. Parameters for Winnt32.exe
Parameter | Function |
---|---|
/checkupgradeonly | Runs a compatibility test on the computer to see whether it has any problems that would interfere with upgrading the OS. Saves a Winnt32.log report in the installation folder for NT upgrades, or an Upgrade.txt report in the Windows folder for Windows 95/98 upgrades. |
/cmd:[command] | Runs the command following the /cmd: parameter after the Windows 2000 Setup Wizard completes. |
/cmdcons | Enables the use of the Recovery Console at boot time for repairing failed installations. Can only be used after Windows 2000 is installed. |
/copydir:[folder name] | Names an additional folder to be copied into the folder in which you install Windows 2000 (\Winnt). The folder remains after Setup completes, and additional folders can be copied by using the parameter multiple times. The folder might contain drivers or other files needed after setup. |
/copysource:[folder name] | Names an additional folder to be copied into the folder in which you install Windows 2000. The folder is deleted after Setup completes. |
/debug[level:filename] | Creates a debug log file with the specified level. The default creates a log file named C:\Winnt32.log with the level set to 2 (Warning). |
/m:[folder name] | Specifies the location of a folder containing system file replacements. Setup checks this folder first for files to copy and then checks the installation folder. |
/makelocalsource | Tells Setup to copy all installation files to the local hard disk so the files will be available later during the installation if the Windows 2000 CD-ROM or network share is inaccessible. |
/noreboot | Tells Setup not to reboot after the initial Windows NT/95/98/2000 file copy phase of Setup is complete. This allows you to run additional commands before continuing. |
/s:[sourcepath] | Specifies the location of the Windows 2000 Setup files. (The default is the current folder.) This must be a full path; for example, X:\path or \\server\share\path. To specify multiple paths in which Setup should search for needed files, use multiple /s: parameters. (You can also sometimes speed transfers from a server by specifying the same source more than once.) Setup fails if the first server isn't available. |
/syspart:[drive letter] | Specifies a hard disk to which you want to copy the Setup startup files. This disk drive is then made active and Setup stops, allowing you to remove the disk and insert it in another computer. When you boot the new computer, the next phase of Setup automatically starts. You have to use the /tempdrive parameter in addition to the /syspart parameter (both pointing to the same drive). |
/tempdrive:[drive letter] | Specifies the drive to which you want Setup to copy its temporary files and install Windows 2000 on. By default, Setup uses the partition with the largest amount of free space. |
/unattend | Upgrades the previous version of Windows in unattended mode, taking all settings from the previous installation. The unattend switch may not be used by original equipment manufacturers (OEMs) selling to end users because it explicitly acknowledges understanding and acceptance of the End-User License Agreement (EULA). |
/unattend:[num:answer file] | Launches Setup in unattended mode by using the answer file you provide. The num parameter specifies the number of seconds to wait after copying files before restarting the computer, and it works only when running Setup from Windows 2000. You must use the /s: switch to specify the location of the answer file. |
/udf:[id,UDF file] | Specifies a uniqueness database file (UDF) to be used to modify an answer file. The ID identifies data in the UDF that should be used in a corresponding section of the ans-wer file. For example, /df:ComputerName,our_company.udf would take the Computer Name from the UDF instead of the answer file. If you don't specify a UDF, you're prompted to insert a disk that contains the $Unique$.udf file. |
As you can see, many of these parameters piggyback onto other parameters, and pretty soon you can find yourself typing (and sometimes retyping) in long strings at the command prompt. If you end up doing this a lot, create a batch file (a text file with the .BAT extension) containing the setup command and parameters. Then simply launch the batch file instead of typing all the parameters.
To use a command-line parameter on a computer without an existing copy of Windows, boot the computer with a Windows 98 boot disk. Then, at the command prompt, type [path]\winnt.exe[parameter], substituting the location of the Windows Setup files for [path]. Table 5-4 shows the available parameters for use only with Winnt.exe, the 16-bit version of Setup.
Table 5-4. Parameters for Winnt.exe
Parameter | Function |
---|---|
/a | Turns on Accessibility features during Setup. |
/e:[command] | Runs the command following the /e: parameter after the Windows 2000 Setup Wizard completes. |
/i:[inf file] | Specifies the filename of the information (.INF) file used for Setup (the default is Dosnet.inf). Do not include the path to the file. |
/r:[folder] | Names an additional folder to be copied into the folder in which you install Windows 2000. The folder remains after Setup completes, and additional folders can be copied by using multiple /r: parameters. |
/rx:[folder] | Names an additional folder to be copied into the folder in which you install Windows 2000. The folder is deleted after Setup completes. |
/s:[sourcepath] | Specifies the location of the Windows 2000 Setup files. (The default is the current folder.) This must be a full path, X:\path, or \\server\share\path. To specify multiple paths for Setup to search for needed files, use multiple /s: parameters. |
/t:[drive letter] | Specifies the drive to which you want Setup to copy its temporary files. By default, Setup uses the partition with the most free space. |
/u:[answer file] | Launches Setup in unattended mode using the answer file you provide. You must use the /s: switch to specify the location of the answer file. |
/udf:[id,UDF file] | Specifies a uniqueness database file (UDF) to be used to modify an answer file. The ID identifies data in the UDF that should be used in a corresponding section of the answer file. For example, /udf:ComputerName,our_company.udf would take the Computer Name from the UDF instead of the answer file. If you don't specify a UDF, you're prompted to insert a disk that contains the $Unique$.udf file. |
A distribution folder (sometimes called a source directory) is a shared folder on a server that contains the \i386 folder (or \ia64 folder for Intel Itanium installations) from the CD-ROM and any device drivers or other files that you add to support the specific systems.
You can create this folder manually, or you can do it while creating an answer file using the Windows Setup Manager (discussed in the next section). To create a distribution folder manually, follow these steps:
With Windows 2000, Windows XP, and the Windows .NET Server family, you can apply service packs to distribution folders. This not only eliminates the service pack step of the setup process, but it also ensures that any time a computer needs original Windows files from the distribution folder, it receives up-to-date files. (So service packs no longer have to be reapplied after changing the system configuration.) To upgrade a distribution folder, see the next section.
Table 5-5. Subfolders you can create to store extra files
Folder | Description |
---|---|
\$OEM$\textmode | The folder in which you place any hardware-dependent files for use while loading Windows 2000 Setup and during the text-mode phase of Setup. These files include any OEM HALs you might use, as well as updated SCSI, keyboard, video, and pointing device drivers. Also include the Txtsetup.oem file in this folder to control the loading and installation of these files. To create the Txtsetup.oem file, create a normal text file and list the filenames of all files in this folder. (Be sure to include this file and all files mentioned in the Unattend.txt file, under the [OEMBootFiles] section.) |
\$OEM$\$$ | The folder that holds any new system files or files that replace existing system files. These files are copied into the various sub-directories that are created in the Windows 2000 system folder (\Winnt). This folder must match exactly the structure of the Windows 2000 system subfolders for those folders in which you want to add or replace system files. For example, to copy new or replacement into files the \%windir%\System32 folder, create an \$OEM$\$$\System32 folder. |
\$OEM$\$1 | The folder in which you place files that you want copied to the drive where Windows 2000 will be installed. Equivalent to the %systemdrive% environment variable and can be used to permit drive letters to be changed without causing problems for applications that point to a hard-coded driver letter. You can also create subfolders here for the files and the entire folder structure will be copied to the system drive. |
\$OEM$\drive_letter | The folder that specifies additional files and folders to be copied into the root folder of the named drive. You will have one entry for each drive having additional files. For example, the files located in the \$OEM$\C folder are copied into the root folder of the C drive during the text-mode phase of Setup. Any subfolders of the \$OEM$\C folder are also copied. All files must use short (8.3) filenames, but you can rename them after installation by including them in the $$Rename.txt file. |
The \Display and \Net folders that were used in the \$OEM$ folder in Windows NT 4 are no longer used in Windows 2000. You can also place the \$OEM$ folder outside of the distribution folder if you place the path (file or Uniform Naming Convention [UNC]) to the \$OEM$ folder after the OEMFILESPATH key in the answer file.
To update a Windows distribution folder to the latest service pack, allowing Windows to be installed with the latest service pack already applied, use the following procedure (you can also integrate hot fixes and security updates into the distribution folder, but it's a nuisance; instead add the hot fixes to an answer file, as described later in this chapter):
Don't update your existing Windows installation share if there are users currently performing installations of Windows using the network share. Bad things might happen, such as failed or corrupted installs that create unstable or nonworking systems.
You can't apply a Windows 2000 service pack to Remote Installation Services CD-based images. To work around this, copy the original Windows CD files to the RemoteInstall folder, apply the service pack, and then add the image to the RIS Server. See Microsoft Knowledge Base article Q258868 for updated information on this issue.
When you run the Update.exe file to update a Windows distribution share with the latest service pack, it creates a Svcpack.log file in the systemroot of the computer you ran Update.exe from. If you want to perform more service pack installations from the same computer, you should rename this log file before performing them.
Microsoft makes operating system hot fixes and security updates available in between service packs. Although hot fixes and security updates are technically different beasts, the way they are deployed is the same, so we'll cover them at the same time. With that said, you should only apply hot fixes on an as-needed basis because they're not regression tested and might introduce system instabilities (make sure you test them out in a nonproduction environment). Security updates should be applied more liberally, although you should still take the time to verify that the update applies to the environment, and test them out before deploying them.
You can download hot fixes and security updates from Microsoft's Security Web site (http://www.microsoft.com/security/), or from Microsoft Corporate Windows Update (http://corporate.windowsupdate.microsoft.com). We recommend that you register with Microsoft to automatically receive security bulletins by e-mail so that you'll be on top of new patches when they come out.
Most companies deploy hot fixes and security updates using Microsoft Systems Management Server (SMS), Group Policy's Software Management And Installation feature (discussed in Chapter 25), or the new Corporate Windows Update service (or the standard Windows Update if you don't mind manually installing on each system). Although you can also integrate hot fixes and security updates into a distribution folder that is integrated with the latest service pack, it's a nuisance (see the Spdeploy.doc file included with the service pack for the procedure). An easier solution is to use the following steps to copy the hot fixes into the distribution folder and add them to an answer file:
Q123456.exe -q -m
Hot fix chaining (installing multiple hot fixes with a single reboot) is supported in Service Pack 3 and later. Previously you had to reboot after each hot fix installation, or run the Qchain.exe tool after installing the hot fixes before rebooting. We should caution you, though, that Microsoft's official recommendation is still to reboot after every hot fix installation, even if you're using post-SP3 hot fixes or Qchain.exe.
Only install hot fixes released after the service pack version you're deploying. For example, if you're deploying Service Pack 3 using an integrated install, only add hot fixes that are labeled sp4 (which means that they're post-Service Pack 3). Installing hot fixes that are included with the service pack breaks the distribution folder.
To configure the distribution folder with additional Plug and Play (PnP) drivers, follow these steps:
\$OEM$\$1\Company\Net
\$OEM$\$1\Company\Video
\$OEM$\$1\Company\Sound
OEMPnPDriversPath = \Drivers.
If you created multiple subfolders for the drivers, add them to the answer file in the following format, each folder reference separated by a semicolon:
OEMPnPDriversPath = Drivers\Nic;Drivers\Video;Drivers\Sound
OemPreinstall = Yes
OEMPnPDriversPath = Drivers\Nic;Drivers\Video;Drivers\Sound
Leave the drive letter out of the paths. The system drive is automatically added to the paths.
net Stop "boot information negotiation layer"
net Start "boot information negotiation layer"
SysPrep and RIS installations postpone installing devices with unsigned drivers until an administrator logs onto the computer. To avert this (something we don't usually recommend), add the DriverSigningPolicy = Ignore line to the [Unattended] section of the answer file. Newer versions of drivers that ship with Windows are installed only if they're signed.
Although it's tempting to add OEM drives to a Remote Installation Preparation (RIPrep)-based image by installing them on the source computer before running the RIPrep program on it, this doesn't work because RIPrep images need to be able to adapt to a wide variety of hardware. Therefore, we recommend that you use the following procedure to add drivers to RIPrep-based images:
[Unattended]
OEMPnPDriversPath = Drivers\Nic;Drivers\Video;Drivers\Sound
net Stop "boot information negotiation layer"
net Start "boot information negotiation layer"
If the hard drive or installation source is attached to a device for which Windows doesn't have drivers, such as a new SCSI adapter or network card, you've got a problem. When performing a manual installation, you can always press F6 at the beginning of text-mode setup to install new mass storage drivers, but this option isn't available with unattended installations. To get around this, you need to copy the mass storage drivers to the distribution folder using the following technique:
d1 = "Windows 2000 Driver Set 1.01", \win2kdsk1, \win2k\ultra160\
to read as follows (the path is replaced with a period):
d1 = "Windows 2000 Driver Set 1.01", \win2kdsk1, .
[MassStorageDrivers]
"scsi controller string from txtsetup.oem" = "OEM"
For example:
[MassStorageDrivers]
"Adaptec Ultra160 Family PCI SCSI Controller (29160, 39160, etc.)" = "OEM"
OemPreinstall = Yes
The Windows 2000 text-mode setup requires the use of MS-DOS 8.3 file formats because the program is based on a 16-bit MS-DOS implementation. Thus, all files that are included in the distribution folder need to have MS-DOS-compliant short names. However, Setup converts them to long names with the use of a renaming file. To create a renaming file, follow these steps:
[media]
filenm1.txt = "Your long filename here.txt"
ding.wav = "Really loud and annoying ding.wav"
whiz.drv = "Whizbang Deluxe Video.drv"
[images]
desktp1.bmp = "corporate logo.bmp"
desktp2.bmp = "division logo.bmp"
When Windows is installed, the setup program pauses several times along the way, waiting for input from the user. Answer files are simple text files that supply the information a user would normally enter, thus automating most of the installation process.
Windows 2000 ships with a handy tool, Setup Manager, which graphically walks you through the creation of answer files. To use this tool, follow these steps:
For real labor savings, set up a typical server or client and then let the Setup Manager Wizard build an answer file that duplicates the configuration of the machine. Then use the answer file to set up multiple versions of the typical machine.
More Info
The Windows Setup Manager walks you through the most important parameters in an answer file, but there are other parameters that might be useful to you. For detailed information on each parameter, open the Deploy.chm file (help file) located in the same Deploy.cab file in which Setup Manager is located.
Figure 5-10. The Windows 2000 Setup Manager Wizard's New Or Existing Answer File screen.
Figure 5-11. The Windows 2000 Setup Manager Wizard's Product To Install screen.
Figure 5-12. The Setup Manager Wizard's User Interaction Level screen.
Real World
Choosing an Interaction Level
When determining the level of user interaction you want to require, you have several choices, and the choice you make determines how much the person running the installation will need to attend to the process. Here's a more detailed explanation of the interaction levels:
Continue entering all the computer names necessary for the number of systems you plan to set up by using this answer file. To remove a name from the list, select the name, and then click Remove. To import a list of names from a text file, click Import. You can select Automatically Generate Computer Names Based On Organization Name to have Setup do it for you—the names will include a few letters from the organization's name but will be otherwise random and unhelpful. When you have finished, click Next.
Figure 5-13. The Setup Manager Wizard's Computer Names screen.
The answer file that the Windows 2000 Setup Manager Wizard creates is an unencrypted text file. If you specify an administrator account password in the answer file, make sure that after installation, you change the administrator account password to something more secure.
Although Windows' default 640 by 480, 60 Hz, 16-color display is pretty horrific, you should use extreme caution here. If you set the screen area (resolution), colors, or refresh frequency too high for some systems, you could damage their monitor or leave them (temporarily) inoperable. We find that 800 by 600, 72 Hz and High Color (16 bit) is a safe setting that's also usable for end users, although if you're sure the hardware is up to it, you might increase the color depth to Full Color (24 bit).
If the systems are going to be joining a domain, choose the Windows Server Domain option. Type the Windows Server domain name in the text box provided. To create an account for the computer in the domain, click the Create Computer Account In The Domain check box and supply a user name and password. The user name and password must be for a user account that has sufficient administrative privileges to authorize the creation of a computer account. Click Next to continue.
If you enter a user name and password with sufficient privileges to create a computer account in the domain for the new systems, you are introducing a security risk because the answer file stores all passwords in a plain text file with no encryption. The best solution is not to include administrative account passwords in an answer file, but if you feel you must, guard the file carefully and make sure you delete it after installation is completed.
Figure 5-14. The Setup Manager Wizard's Distribution Folder Name screen.
Setup Manager, by default, creates a distribution folder named \Win2000dist. Unfortunately, it doesn't create an \i386 subfolder, which causes integrated service pack installations to fail. Therefore, we recommend that you create an \i386 folder inside the distribution folder, and then move the contents of the distribution folder into the new \i386 folder.
If you have additional drivers for Plug and Play devices that aren't included with Windows 2000, copy them to the \Plug and Play Drivers folder on the system drive. Windows 2000 Setup searches this folder for drivers it couldn't locate.
Figure 5-15. The Setup Manager Wizard's Copy Additional Files Or Folders screen.
If you created an answer file automating a CD-based installation, rename the Unattend.txt file Winnt.sif, place it on a floppy disk, and insert the floppy disk right after the computer boots from the Windows 2000 CD.
The batch files that Windows 2000 Setup Manager Wizard creates use the 32-bit version of Setup. If you don't have a 32-bit version of Windows on the machine where you're planning to install Windows 2000, modify the batch file to use the 16-bit Setup file by deleting the 32 from the third to the last line of the file. Also change the /unattend: parameter in the second to last line to /u:.
Although creating answer files for running Windows 2000 Setup in an automated fashion is very convenient and can save a lot of time for administrators, you can go a step or two further than that under the right circumstances.
One way to further automate the setup process for a new machine is to "clone" an existing system. (One can also use Remote Installation Services in combination with IntelliMirror to achieve similar results, as discussed in Chapter 25.) The primary advantage of cloning a system is installation speed: a SysPrep-created disk image can usually be installed 45 to 60 minutes quicker than a normal installation.
Cloning works like this: First, install Windows and all of the applications you need on a single machine that is identical or very similar to the many machines on which you want to deploy Windows. Then prepare this system for cloning using SysPrep (which is available on the Windows CD) to clear out the SID and other computer identity information. Clone the configuration using a third-party disk-imaging program, such as PowerQuest's Drive Image or Norton Ghost, which copies and compresses the disk image to a network share. You can then boot up a blank system using the floppy disk created by the disk-imaging program, copy and uncompress the cloned disk image onto the new system, and be up and running in far less time than would be required to perform even a fully automated operating system and application installation.
However, this solution has some problems. First, the systems need to be fairly similar for the disk images to work on them. The computers don't have to be exactly the same because Windows leverages Plug and Play to detect changes to most system components. However, the systems do need to have identical mass storage controllers (SCSI controller or IDE chipset), as well as share the same HAL—no mixing ACPI systems with non-ACPI systems or uniprocessor systems with multiprocessor systems. Also, this process doesn't work for domain controllers, unless you script the DCPROMO process into the disk image.
Covering in detail how to use SysPrep and disk-imaging tools to mass deploy preconfigured computers is outside the scope of this book, but we can give you a taste of how it works. Thoroughly investigate these tools and experiment with them in a test environment before installing Windows throughout your organization using this procedure. The following steps summarize how to clone a system:
Because a SysPrep-created operating system image can't be installed on a disk partition smaller than that on which the image was created, we recommend that you install Windows into a partition just large enough for Windows and any necessary applications. The destination computer's hard disk will be partitioned to the same size as the source computer's partition, but you can use the ExtendOemPartition entry in the Sysprep.inf file to extend the destination computer's system partition to fill the hard disk.
You can optionally provide an answer file to modify individual machines' system configurations without re-creating the disk image. To do this, create an Unattend.txt file as you would normally, rename it Sysprep.inf, place it on a floppy disk, and insert the floppy disk right before the mini-Setup Wizard is launched.