The computers on which you are installing an operating system must all be connected to a reliable, high speed network. Wide area networks (WANs) that are connected with analog modems or Integrated Services Digital Network (ISDN) do not have sufficient bandwidth to support RIS installation.

To take full advantage of RIS, each of the computers on which you are installing an operating system must have a Peripheral Component Interconnect (PCI), Mini-PCI, or CardBus type of network adapter that supports the PXE specification. PXE-enabled network adapters allow client computers to be remotely booted. In addition, the basic input/output system (BIOS) for each computer must be capable of using the network adapter as a boot device (also known as remote-boot capability). A ROM BIOS that is at least version .99n satisfies this requirement. Also, most computers that conform to the Net PC or PC98 specifications have a PXE remote boot–enabled network adapter and remote boot–enabled BIOS.

The published PXE specification defines the remote boot process and also establishes the PXE compliance standards for hardware manufacturers and other vendors. RIS uses PXE environment extensions to Dynamic Host Configuration Protocol (DHCP), an industry-supported technology, to allow workstations to do the following:

	Boot remotely using their network adapters to access boot code from a network location.
	Install an operating system from a remote source to a client’s local hard disk.

The PXE environment is built upon Internet protocols and services that are widely used in the computer industry. This includes TCP/IP, DHCP, and TFTP. The PXE extensions to the DHCP protocol enable information to be sent to network-bootable systems and also allow these systems to locate remote installation services.

Your network must support TCP/IP to perform RIS installations.

You need to install Remote Installation Services on a dedicated server running Windows Server 2003. The RIS server must have at least two NTFS disk partitions — one for booting the server operating system and another to contain the directory structure for the client operating system images or distribution shares. The partition containing the images or distribution shares must be large enough to store one or more operating system images, depending on your requirements. In addition, the RIS server must run the following services: Remote Installation Services, Single Instance Store service (SIS), and the Trivial File Transfer Protocol Daemon (TFTPD).

RIS servers rely on the Domain Name System (DNS) to locate the required Active Directory domain controllers to facilitate domain operations. If you use Windows Server 2003 DNS, you have the benefit of dynamic updates for your DNS server. However, it is not a requirement to use Windows Server 2003 DNS for RIS to function. Whichever DNS server software you use, it must support the SRV RR record type and the dynamic update protocol specified in RFCs 2052 and 2136, respectively.

RIS servers require a DHCP server on the network which is authorized and has an activated scope. Remote boot–enabled computers must receive an IP address from a DHCP server before they can contact a RIS server to request an operating system installation. You can install Windows Server 2003 DHCP or you can use the DHCP services provided with Windows 2000 Server. In addition, you can use a non-Microsoft DHCP server software.

You must install RIS on a computer running Windows Server 2003 in an Active Directory domain. For best results, configure this computer as a member server. Although you can install RIS on a domain controller, the heavy traffic load generated by RIS can impact the performance of the domain controller.

RIS uses Active Directory to locate RIS clients and other RIS servers. You can administer the RIS server from the Active Directory Users and Computers snap-in (Dsa.msc) located on the RIS server. For more information about Active Directory, see the Directory Services Guide of the Windows Server 2003 Resource Kit.