227 Chapter 16: Installing and Setting Up Windows Server 2008 N Disk A fast SCSI-based disk subsystem is important, particularly for servers that will store a lot of data. (See Chapter 13 for more information about choosing SCSI systems, using different RAID levels, and other important disk information.) Use the information in Chapter 13 to help you size your server, but remember this rule of thumb: Get the most capable server you can afford and make sure it is expandable to meet your future needs, through the addition of more RAM, more processors, and more disk space. Even with all of that, it is common for servers to be replaced three to four years from the date they were placed into service. Testing the Server Hardware You found all your server hardware in the Windows Server 2008 HCL, you made sure your server is adequately sized, you purchased it, and you have your shiny new Windows Server 2008 DVD-ROMs sitting there, all ready to be installed. Is it time to start the installation yet? Well, not quite. Before installing any network operating system (NOS), particularly on a server that will be used for production, make sure you carry out hardware testing (also called burn-in) on the server before installing Windows Server 2008. Computer hardware tends to be most reliable after it has been running for a while. In other words, failures tend to happen when equipment is new, and the chance of hardware failure decreases rapidly after the hardware has been up and running for 30 to 90 days. Because of this, it‘s a good idea to test new servers for at least a week (testing for two weeks is even better) before proceeding to install the NOS. Doing this can help provoke any early failures in the equipment during a time when they’re easy to fix and they won’t affect any users or the network. Moreover, many servers have a 30-day return or exchange policy from their manufacturer, so if you discover problems, you’ll have a chance to return the system and perhaps start over with a different model. You test the hardware using diagnostic software that came with the server computer or is available from the maker of the server. Most such diagnostic software lets you to choose which components of the system are tested and enables you to test them in an endless loop, logging any discovered errors to a floppy disk, USB key, or the screen. You should focus the tests on the following components: N Processor(s) N System board components, such as interrupt controllers, direct memory access (DMA) controllers, and other motherboard support circuitry N RAM N Disk surfaces 228 Networking: A Beginner’s Guide TIP Server-testing software often enables you to choose between nondestructive and destructive testing of the disks. (Destructive means any data on the disks is erased during the testing.) Destructive testing is best to discover any errors on the disks. This is one reason that you want to carry out this testing before you install your NOS. If the diagnostic software allows you to do so, you can usually safely skip testing components such as the keyboard or the display. Your primary concern is that the unit continues running properly when it is under load for an extended period of time. You also want to make sure that the RAM is working properly and that no bad sectors show up on the disks during testing. It’s also a good idea during testing to power the unit on and shut it down a number of times, since the impact to the unit of initially powering on often can provoke a failure in any marginal components, especially if the unit is allowed to cool down first. Surveying the Server Prior to an In-Place Upgrade The Windows Server 2008 family of products takes advantage of plug and play (PnP) hardware and can detect and automatically configure any PnP devices to work with Windows Server 2008 during the installation. PnP is not perfect, though. For one thing, you might have installed components that are not PnP devices, and Windows Server 2008 will not be able to configure those devices. Also, sometimes PnP devices can conflict with other devices, or the drivers for a specific device might not allow proper configuration for some reason. Because of these imperfections, it’s important to survey the components installed in the server before installing Windows Server 2008 as an upgrade. Performing a survey is not really important when setting up a new server. For the survey, write down all the installed devices, along with the resources that each one uses in the server. The resources include the IRQ channel, DMA channel, and memory I/O addresses used by each device. Then, if a device isn’t working properly after you install Windows Server 2008, you might be able to configure the device manually to known settings that work. NOTE Some server computers come with utilities such as HP’s SmartStart. Such utilities handle the server at a hardware level and keep the information in a space separate from the NOS. Server utilities such as HP’s make life much easier when you are trying to troubleshoot a hardware problem with the server. Making Preinstallation Decisions After configuring, checking, preparing, and testing your hardware, you can actually begin installing Windows Server 2008. During this process, you first spend time making a number of important preinstallation decisions that you must be prepared to specify during the installation. The following sections discuss these choices. 229 Chapter 16: Installing and Setting Up Windows Server 2008 NOTE This chapter and the two following chapters provide an overall introduction to Windows Server 2008. Certain advanced installation scenarios and techniques are not described here. To learn about other features and choices available when installing, administering, or using Windows Server 2008, consult a book devoted to Windows Server 2008, such as Microsoft Windows Server 2008: The Complete Reference by Danielle Ruest and Nelson Ruest (McGraw-Hill/ Professional, 2008). Upgrade or Install? You can upgrade a server running Windows Server 2003 to Windows Server 2008 or perform a full installation, where you wipe out any existing NOS on the server. The main benefit to upgrading is that all your existing settings under Windows Server 2003 will be maintained and automatically carried forward into your Windows Server 2008 installation. These include networking details, such as TCP/IP configuration information, as well as security settings that you might have tediously set up over time. In fact, if the server can be upgraded, you should plan on doing so, unless you need to change something fundamental in the server, such as changing the disk format from FAT32 to NTFS. NOTE Different upgrade paths exist depending on which edition of Windows Server 2003 you were running. You can upgrade Windows Server 2003 Standard Edition to 2008 Standard or 2008 Enterprise. You can upgrade Windows Server 2003 Enterprise Edition to 2008 Enterprise or 2008 Datacenter Edition. FAT or NTFS? Windows Server 2008 supports hard disks formatted using either File Allocation Table (FAT16 and FAT32) or NT File System (NTFS). NTFS is required for any Windows Server 2008 servers that will function as domain controllers. NTFS also is the only file system that enables you to take full advantage of Windows Server 2008’s security features. Moreover, NTFS is optimized for server performance and performs better than FAT under almost all circumstances. Domain Controller, Member Server, or Stand-Alone Server? Another choice you need to make is in which mode you will configure your server. To make this decision, you need to understand two important concepts: Windows Server 2008 domains and workgroups. A domain is a sophisticated administrative grouping of computers on a Windows network that makes it possible to administer the network’s resources from a single point and to implement strong security. Domains enable you to manage multiple Windows Server 2008 or Windows 2003 servers more easily. A workgroup is a simple collection of computers on a network and is suited only to pure peer-to-peer networks. 230 Networking: A Beginner’s Guide You can configure Windows Server 2008 in one of three modes to support either domains or workgroups, as follows: N Domain controllers hold the domain’s Active Directory information and authenticate users and access to resources. Most Windows Server 2008 networks have at least one domain and therefore need at least one domain controller. N Member servers are part of a domain, but do not hold a copy of the Active Directory information. N Stand-alone servers do not participate in a domain, but instead participate in a workgroup. Earlier versions of Windows servers (NT and 2000) needed to be designated as either primary domain controllers (PDCs) or backup domain controllers (BDCs). The PDC performed all administrative tasks, and the BDCs simply kept read-only copies of the domain information to continue authenticating security on the network in case the PDC failed. Newer Windows servers, such as those running Windows Server 2008, simplify matters, so that all Windows Server 2008 domain controllers are just that—domain controllers. Each domain controller holds a copy of the Active Directory data and can perform all the functions of the other domain controllers. Windows Server 2008 uses the concept of multimaster domain controllers, which all seamlessly operate the same way as the other domain controllers. TIP Except in the smallest of networks, it’s a very good idea to have two domain controllers. This way, all of your domain information is preserved and available to the network should one of the domain controllers crash. Domain information is automatically synchronized between the available domain controllers. Per Seat or Per Server? Yet another important choice to make when installing Windows Server 2008 is how the server will manage its Client Access Licenses (CALs). Windows Server 2008 supports two different ways of managing CALs: N Per-server licensing assigns the CALs to the server, which will allow only as many connections from computers as there are installed CALs on that server. N Per-seat licensing requires purchasing a CAL for each of your client computers, which gives them the right to access as many Windows servers as they wish; the servers will not monitor the number of connections. 231 Chapter 16: Installing and Setting Up Windows Server 2008 Generally, Microsoft recommends that you use per-server licensing when running a single server and per-seat licensing when running multiple servers. If you are unsure of which mode to use, Microsoft recommends that you choose per server, because Microsoft lets you change to per-seat mode once at no cost (changing from per seat to per server has a price). Carefully review licensing options with your Windows Server 2008 reseller to determine the most economical way to license your network servers properly. Wait! Back Up Before Upgrading! If you are installing Windows Server 2008 as an upgrade to another NOS, such as Windows Server 2003, it’s vital that you fully back up the server prior to installing Windows Server 2008. (It’s a good idea to make two identical backups, just in case.) You should use whatever backup software you normally use for your existing NOS, making sure the software can properly restore the previous NOS in case you need to “unwind” the upgrade process and revert to your starting point. Even when you are performing an upgrade and will not be reformatting any of the disks, making a preinstallation backup is good insurance in case of trouble. Installing Windows Server 2008 To begin the installation of Windows Server 2008, you can either configure the server computer to boot from the Windows Server 2008 DVD-ROM or insert the Windows Server 2008 DVD-ROM while running Windows Server 2003. Most servers can boot from their DVD-ROM drives, which is the best way to start the installation. Running the Windows Server 2008 Setup Program The following steps outline the process of running the installation program for Windows Server 2008 and installing it onto a server. If you are learning about Windows Server 2008 and have a suitable computer to use, you should take the time to install Windows Server 2008 so that you understand how the process works. Or, if you like, you can read along through the following steps to familiarize yourself with the installation process. (I recommend actually performing an installation such as the one described here, and then playing with the resulting server as a way of more quickly and completely learning about Windows Server 2008.) 1. When you boot from the Windows Server 2008 DVD-ROM, the program first presents a screen that lets you choose the language to install, the formatting to use for time and currency displays, and the keyboard or input method (see Figure 16-1). Make the appropriate choices, and then click Next to continue. . 2008 family of products takes advantage of plug and play (PnP) hardware and can detect and automatically configure any PnP devices to work with Windows Server 2008 during the installation the chance of hardware failure decreases rapidly after the hardware has been up and running for 30 to 90 days. Because of this, it‘s a good idea to test new servers for at least a week (testing. chapters provide an overall introduction to Windows Server 2008. Certain advanced installation scenarios and techniques are not described here. To learn about other features and choices available