Remote Installation Services tips and tricks
By Jason Rush, Technical Writer, Microsoft Corp.
You use Remote Installation Services to install Windows 2000 onto remote computers over the network. The Windows 2000 source image can be customized, and can include applications.
Use the appropriate number of Remote Installation Services servers on your network.
In a small local area network (LAN) --for example, one physical subnet without a router--a single Remote Installation Services server can serve all PXE remote boot-enabled client computers up to the network bandwidth, or server resource limitations. In branch office situations, where only slow links exist to the branch site, it may be worthwhile to physically locate a Remote Installation Services server at the branch site to avoid saturation of the slow link to that site. Because the PXE remote boot process is based on DHCP, routers between Remote Installation Services clients and Remote Installation Services servers must be configured to forward DHCP broadcasts in order for the client requests to be received and answered by the RIS server.
Restrict the number of installation options and operating system choices a user has access to within the Client Installation wizard.
By restricting the installation options, you increase the percentage of successful operating system installations without requiring assistance from technical support or administrative staff. By default, Remote Installation Services ships with only one installation option and operating system choice (both are selected by default without prompting users).
Use the Remote Installation Preparation wizard (RIPrep) image format to deploy a Windows 2000 Professional standard corporate desktop configuration across different types of client hardware throughout your organization.
Using the Remote Installation Preparation wizard, an administrator can replicate the installation image of an existing Windows 2000 Professional client computer, including locally installed applications and operating system configuration changes, to an available remote installation server on the network. After it is replicated, the installation image can be remotely installed by any supported client computer that uses the same hardware abstraction layer (HAL) as the source computer, regardless of other hardware differences between the source computer used to create the image and the destination computer installing that image.
Verify the versions of PXE ROMs in use on client computers.
Remote Installation Services requires a minimum version of .99L PXE ROMs in client computers to function correctly in all situations. Depending on the network interface card (NIC) brand and model and the implementation of RIS servers in your organization, PXE ROM revisions starting at .99C may work. If you experience difficulty, check the version of the PXE ROM in use, and if necessary contact the NIC or system vendor for an update.
Use Remote Installation Services with computers that do not contain the PXE-based remote boot ROM.
Use the remote boot floppy generator (Rbfg.exe) to create a floppy disk for computers that do not contain the PXE-based remote boot ROM. You can then use the Remote Installation Services feature for these computers.
Avoid non-ASCII characters when using the Client Installation wizard.
Limit the characters to only standard ASCII characters (OEM characters 32-127) for user name, password, or domain name information. The Client Installation wizard does not support extended ASCII character sets (such as those containing �, �, and so on).
Do not install the DHCP service only to authorize a RIS server.
If the RIS server is not going to be a DHCP server, do not install the DHCP service to authorize the RIS server. If the DHCP Management snap-in is only needed to authorize the RIS server, then install this snap-in by running adminpak.msi from %windir%system32.
Plan carefully when placing DHCP and RIS on the same server.
This may work out well in smaller environments but if you add another RIS server it will not be used, since PXE clients will get both IP and RIS server info from the existing servers. In other words, the original DHCP/RIS servers will continue to be overloaded.
This was first published in January 2001