With the release of Windows Server 2008, it is a good idea now to start to practice and learn how to install 2008, implement needed functionality (such as roles and other features) and analyze the new installation options that are now available with 2008 such as Core Server installation, using Windows Deployment Services (WDS) as well as unattended installations, upgrades and/or dual boot scenarios. This article will cover the fundamentals of installing a full copy of Windows Server 2008 on enterprise class server hardware.
With Windows Server 2008, you will find new tools to deploy your server. With 2008 you will find that not only is it easier to install a Windows Server, but now you have more options that fit your production needs and you can now install (or not install) what it is you want or need. For example, if you wanted to be selective and run a server with the most minimal set of needed services, then you would select the Server Core option when running your installation. If you wanted to deploy Windows Server 2008 with an unattended answer file, you can with new options and features, to make your deployment easier.
In this article we cover the basic installation options available to you when deploying a full copy of Windows Server 2008 on an enterprise class server. The article’s goal is to familiarize and acclimate you to the process of deploying Windows Server 2008 on systems on your production network.
It is important to note that although 2008 is a fantastic product with many new features and improved functionality, you should never run a beta or test copy of Windows in your production environment. It is not supported nor is it a complete product. As well, you should never rollout any new operating systems (or run upgrades) on production systems or networks without first testing the official release in a test lab to ensure that your applications are compatible, your drivers, your hardware and so on.
The first step in deploying a Windows Server 2008 system is to do all of the analysis needed upfront to appropriately size your server hardware and prepare for an easy installation. There is not enough that can be said about ‘pre-planning’ your installation. Make sure that you have all the software you need, appropriate drivers and so on. You should create a checklist or some form of list to help you keep track of what you need, what you need to do and what steps have been accomplished. This is helpful for problems that you may encounter during the installation – you can quickly ascertain the source of the problem when you have a checklist to follow so you can retrace your steps. You should always pre-plan any deployment (no matter how big or small) and run thorough tests so that you know what works and what does not before installing new systems on your network or upgrading older systems as an example. For another example, when we first started to run Windows Server 2008 in its earliest forms (Longhorn) for testing purposes, it was quickly found that it was extremely difficult to install on an enterprise class server due to the fact that most of the enterprise server hardware vendors had yet to create alternatives to the new release of Windows and most Windows Server 2003 drivers would (and will) not function on the newest version of Windows, 2008.
One of the biggest earlier issues when testing Windows Server 2008 was its incompatible RAID drivers that just ceased the installation completely for lack of support and would cease the ability to continue testing. Some of the ways around this were to virtualize the instance, although this is not going to fit into everyone’s deployment schedule and needs therefore testing on the enterprise class equipment became paramount in our (and other testers) plans. Planning your deployment when working with an enterprise class server is very important, not planning will cause you to deal with problems as they occur and waste your time (such as hunting down updated drivers). When working with high end equipment, it is common to work directly with the vendors support group to get newly developed drivers that are supported under Windows Server 2008. Just as many of us (and I am sure many of you) were testing the beta version of Windows, it was wise to check with HP, Dell and other hardware vendors to see what the availability was (and is) for updated firmware and software from the makers. Some even offered beta assistance which also proved helpful in testing Windows Server 2008 on a RAID array.
So, why so much fuss about a high end system – or one called an ‘enterprise-class’ system? Simply put, most data centers run their systems on server-class hardware, not PC based hardware. An enterprise class server is a system that is built for large networks and is usually extremely powerful, scalable and redundant. Although it costs more, it gives you more functionality and flexibility when deploying and is extremely easy to fix when problems do arise – the servers can even be kept running without interruption while failed components are replaced and/or fixed.
An enterprise class server is able to be scaled up to meet the demands of enterprise class applications, such as SQL Server, Exchange, and application-based middleware. An enterprise class server usually contains advanced hardware and many times, most of these server systems (Dell, HP, IBM, etc) come with their own software tool suites which are used to install drivers and management software into Windows so that Windows can interoperate with the enterprise class hardware. An example of a high end enterprise class system is the HP DL380 as seen in Figure 1.
Figure 1: An Enterprise Class Server System
In Figure 1, you can see that most, if not all hardware installed on the server is redundant. Redundant power supplies, 2 CPU Slots (one populated), fans and network connections allow for complete failover capabilities and most if not all of these FRUs are modular, hot swappable and built to keep the server running no matter what issue arises. Match this up with redundant UPS connections and you can possible reach 5 nines (99.999) uptime if you do not need to shut down the system much, or if just running a few updates, hotfixes and service packs as needed.
One of the main issues you may encounter when installing Windows Server 2008 comes in the form of installing RAID drivers and finding ones that are supported by Windows Server 2008 in its pre-release form. Since the inception of NT 3.x and 4.0, this process has never been easy. In earlier forms of the operating system (as previously mentioned - NT as an example), you had to hit a specific key (F6) to add RAID drivers while installing, force drivers in if not available and fight with the system to get it to function correctly and remain stable. A lot has changed since, but one thing that will never change is when new operating systems come out, the vendors have to keep up and create software to run on the new platform and that process usually takes time and usually never comes out during the beta process of any new release underway – very rarely do you receive fully functional, tested and supported drivers until a full release. That being said, it is possible to create a RAID array if you can find the drivers and a system that will allow for it until the final release of Windows Server 2008 is official, by then just about every operational hardware vendor will have drivers, software and other tools ready to go for 2008.
You should make sure that any software tools, firmware and drivers you download and use in their early forms are checked again (and updated if needed) as Windows Server 2008 officially hits the market.
RAID is extremely important to consider and implement when deploying an enterprise class server. Since redundancy is important due to unexpected issues (such as a drive reaching its MTBF) which causes the drive to fail and causes the server to crash, you should consider this when deploying a new server. RAID is nothing more than a Redundant Array of Inexpensive Disks. Figure 2 shows an example of an Enterprise Class server configured as an array with 6 SCSI hard disks. Figure 3 shows an external RAID array that can be connected via Fiber or SCSI.
Figure 2: Viewing a RAID Array (Internal)
Figure 3: Viewing a RAID Array (External)
RAID comes in many levels. You can configure RAID in many ways. Table 1 lists a chart of the most commonly used RAID levels and their fault tolerance (or lack thereof).
Disks Needed to Implement
Striping + Duplexing
Striping with Parity
Duplexing and Mirroring are similar in that they both use two disks and each carry a copy of the data, the main difference lies in the disk controller. When two disk controllers are used (one for each disk), this is duplexing. This also increases fault tolerance. When mirroring is used there is only one controller. Although you have two disks in use, your single point of failure becomes the controller itself.
Striping is commonly used for an increase in performance whereas parity is used to add ‘redundancy’ to your solution. Parity commonly increases your need for disk space to carry a copy of the parity data.
Although there are many other RAID levels, these are absolutely the most commonly used today. You can also mix and match RAID levels based on how you create your partitions. You must know the available levels and know which level(s) you want to use when you deploy.
Once you have selected the enterprise class hardware you wish to use, you must make sure that you have all of the drivers and software needed for deployment. As 2008 is fairly new (just getting ready for prime time) as mentioned before, it’s likely that you will not have full driver support, or in some cases, the management software used for the server, may not run on 2008. Because of this, it’s important to make sure that you have visited Microsoft’s Web site to check for the current drivers and software and then check your server vendors Web site to check for any updates they have as well. You can also discuss options about the server hardware in a pre or post sales vendor meeting or contact them directly though their support centers. In either case, you will find yourself in a better position when you have planned out your install in advance, have gathered all of the needed software and drivers and prepped the server correctly for install.
Always use RAID – Disks are meant to fail. The MTBF (which stands for mean time between failures) is a reliability rating method used to indicate the expected failure rate of a product in power on hours (POH). This being said, drives ‘will’ fail and you absolutely need to make sure that you have a way to resolve a failed drive when in fact it does.
When deploying Windows Server 2008, it’s important to consider and adhere to the minimum requirements needed for the system to be installed. If you do not meet the bare minimum requirements, the installation will not continue. Table 2 shows a list of minimum and maximum system requirement levels for your review.
1 GHz (for x86 processors) or 1.4 GHz (for x64 processors)
2 GHz or faster
2 GB or more
Hard Disk (System Partition)
10 GB Free Space
40 GB or more
Super VGA (800 x 600) or higher-resolution monitor
Super VGA (800 x 600) or higher-resolution monitor
Keyboard and mouse (or other compatible pointing device)
Keyboard and mouse (or other compatible pointing device)
Table 2: Windows Server 2008 System Requirements
When sizing your installation, you should always consider the pre-planning discussed earlier. If you have a checklist or worksheet that shows you the processing power of the combined applications on the server that will be deployed, you may find that you need to add CPU’s, or increase their power as an example. Other confusing items may be dual core CPU’s. It should be noted that CPU performance depends not only on the clock frequency of the processor, but also on the number of processor cores and the size of the processor cache. An Intel Itanium 2 processor is required for Windows Server 2008 for Itanium-Based Systems as an example. Now that you know what you need, let’s begin deploying Windows Server 2008.
Configuring your Hardware
To install Windows Server 2008 correctly, you must first configure your hardware. When working with enterprise class server systems such as the HP DL380, you will want to configure your storage correctly, so that you can install Windows Server 2008 within the needed hard disk space requirements. To do this, you may need to configure RAID. To configure RAID, you will likely have to use your servers BIOS, or the management utility that came with it. In this example, we will use the system BIOS.
Figure 4: Configuring a RAID Array
Earlier we mentioned that you can have multiple RAID levels exist on the same Server. Figure 5 shows the BIOS configuration for an HP DL380 being prepared for an installation of Windows Server 2008. Here in this example we are configuring the HP Smart Array 6i which comes with its own controller, or HBA (Host Bus Adapter). In figure 5, you can configure the drives for redundancy.
Figure 5: Configuring Logical Drives in the Array
Once RAID is configured and setup, you can now install Windows Server 2008 on the newly prepared hardware.
Always use RAID if possible. In the previous example, the drives were configured so that a system crash or hardware failure could be corrected through the use of RAID. For the previous example, the disks in which the operating system is installed are mirrored in a RAID 0+1 configuration. The remaining disks are configured to support RAID 5 with a hot spare. In this type of configuration you can avert disaster by having a backup for your mirrored system drive, and a backup for your data located on the second RAID configuration. In the BIOS it will show up as two logical drives (one with almost 70 GB of space) and the 2nd drive with over 200 GB of space. Now, you can install Windows without issue, with plenty of usable space ‘and’ if you experience any kind of disk failure, it is likely that you can quickly fix and resolve the issue with this type of configuration.
Manually Installing Windows Server 2008
A Windows Server 2008 manual installation is fairly easy. Once you have all of the pre-planning and pre-requisites established all you need to do now is run the installation program and get started. The checklist or worksheet that was recommended should point you in the direction of getting the disks and/or software ready for the installation process. You should view and check the integrity of your disks before the installation process begins.
When installing Windows Server 2008, you will find that setup works in several stages:
- First, you put in the installation media (if installing from CD/DVD-ROM) and then running the installation program if auto-install does not kick off. Figure 6 shows the initial dialog to install Windows Server 2008. Click next and begin the installation. Once you choose to ‘Install Now’, you will be given the option to configure and enter the activation key.
Figure 6: Installing Windows Server 2008
- Next, enter the activation key and click - Next. If you do not have a key, you will not be able to use Windows Server 2008 at all.
- Once you have entered the key, you will now be given a choice on what type of installation you would like to begin. Figure 7 shows that you can select a full installation (which we will do here), or a Server Core, which only installs the basic services and functionality needed, not the complete system. The two options are:
- Windows Server 2008 (Full Installation): This option installs the complete installation of Windows Server 2008. This installation includes the entire user interface, and it supports all of the server roles.
- Windows Server 2008 (Server Core Installation): This option installs a minimal server installation of Windows Server 2008, which you can use to run supported server roles through the command prompt interface.
Figure 7: Picking a Full Installation or Server Core
- Next, agree to the license terms and click Next to continue. If you do not agree, you will not be able to install Windows Server 2008.
- Next, you can agree to an installation, or if the installation program detects a pervious installation of Windows, it will ask you for a standard upgrade or custom installation which will install a fresh copy of Windows Server 2008 on your system.
- Next you can configure drive options. If you select ‘Advanced’, you can configure the specifics of your drives and partitions and how you want Windows to layout the system on your drives.
You must ensure that when you select the option to do an upgrade that you have in fact checked what you can (or can’t upgrade to and from). With Windows Server 2008 having been released to manufacturing and on its track to worldwide availability, one aspect of the transition to the new server platform concerns the upgrade paths from Windows Server 2003 which is what most customers are currently running. If you are running Windows Server 2003 in your enterprise and using 2003 based ADS, DNS, and DHCP and so on, you are the ‘best’ candidate for an upgrade.
It’s been recommended by Microsoft that if you aren’t running an ‘ideal’ or ‘optimum’ scenario such as this, you are best served with a ‘clean installation’ scenario. In this case, you will have to backup all of your production data, install (and upgrade) to Windows Server 2008, test thoroughly and then add data and ensure that everything is running properly and as expected. As mentioned earlier in this article, it is extremely wise to test all of your software, drivers, firmware, applications and programs ‘before’ installing or upgrade to 2008. That being said, if you do in fact upgrade (or install fresh), you will know that your applications work in advance and will not cause you problems when moving into production.
- Now, the files will install on your system and the installation will commence.
- Once the installation has been completed, you will now be ready to log on and use Windows Server 2008 for the first time as seen in Figure 8.
Figure 8: Logging into Windows Server 2008 for the First Time
Now that you have a completely installed Windows Server 2008 system running on enterprise-class hardware and entered into production… you should now check your system logs and ensure that you do in fact have everything running correctly and as expected. It also helps to now get a baseline of the system and see how it operates under no load, expected load and then do periodic and scheduled checks as you continue to use your system to ensure that it in fact is operating as expected.
Installing a Server Core
Although this article’s focus is not on the Server Core installation, it should be mentioned and links are provided for extra information if you need it. When installing a Server Core installation of Windows Server 2008 on an x86-based or x64-based server, you will be prompted during Setup to install the most minimal set of requirements needed for basic (or purpose driven) functionality and extreme security. This option installs a minimal server installation of Windows Server 2008, which you can use to run supported server roles through the command prompt interface and will only give you exactly what it is you ask for. This is very reminiscent to the says of Novell NetWare, and today’s Unix and Linux based systems.
When you select this option, Setup will install the files that are required for the selected server roles to function only – nothing more will be installed without your knowledge or decision.
As an example, the GUI which we all know as the windows interface will not be installed—you configure and manage the server locally from the command prompt only and will not be given an option to work within the GUI because it simply does not exist! You should consider this option because it will reduce the servicing and management requirements and the overall attack surface of your system which in turn helps to enforce your security policy and so on.
There are other installations methods that will be covered in future articles and can also be found below in the link section of this article. You can run a Server Core installation which only installs the bare minimum functionality needed, unattended setups and multiple installations simultaneously. It is recommended that no matter what installation method you choose, you spend quality time beforehand developing pre-planning steps, designs, and any other checklists/worksheets that can help you deploy the system with ease.
Once you have your Windows Server 2008 system installed and configured, you may encounter a few issues. To resolve them, simply take note of the issue (any error codes, screenshots and warning messages) that has occurred and look it up online, in books or try to recreate on a non-production test-lab system.
Troubleshooting Your Installation
When installing Windows Server 2008, you may run into issues that may need your attention. As mentioned earlier, if you do not (or did not) get any updated drivers from the vendors, or providers of your systems and service – its likely that you will encounter installation problems with your system. Even if you have all of what you believe you need, you could still encounter issues. Some unexpected issues you may encounter include the following list. Some of the more ‘obscure’ issues you may encounter include (but not limited to):
- Incorrect, unusable or unsupported drivers, firmware, hardware, software (applications, programs, etc) and lack of vendor updates, patches and fixes.
- Unsupported file systems such as FAT.
- Loss of power while installing – if over network, loss of connectivity while running installation.
- Corrupted installation media. If using a DVD-ROM as an example, the installation media may become corrupted. Sometimes when installing an ISO to a disk for installation, the burn speed (if too high as en example) can cause an unusable disk.
- Any error messages that you cannot decrypt need to be looked up on the Microsoft Support site and Knowledge base for further clarification.
If you are having problems with an ‘upgrade’, make sure that you are following the correct upgrade paths posted by Microsoft:
- Users of Windows Server 2003 Standard Edition (R2, SP1 or SP2) can perform a full installation of Windows Server 2008 Standard Edition or Enterprise Edition.
- From Windows Server 2003 Enterprise Edition (R2, SP1 or SP2) you can upgrade to a full installation of Windows Server 2008 Enterprise Edition
- From Windows Server 2003 Datacenter Edition (R2, SP1 or SP2), you can upgrade to a full installation of Windows Server 2008 Datacenter Edition.
In this article we covered the fundamental concepts of installing Windows Server 2008 on enterprise class servers and hardware such as initial preparation, deployment steps and caveats to watch out for. In this article we learned the basics of installing Windows Server 2008 (full installation) on hardware, configuring RAID and ensuring that the installation went smoothly. To learn more about Windows Server 2008, Installations and other helpful tips, check out the rest of the site and the links section below.
Windows Deployment Services
- Server roles available in Server Core installations
- Options for Server Core Installation
- 2008 Installation Readme