Legacy servers and their ancient applications are ideal candidates for virtualization, but a virtual server migration...
to new hardware brings several challenges.
Virtualizing a legacy OS on newer hardware can make the system more efficient, reliable and cost-effective. But virtualization hardware supportability, device drivers, and the virtual server migration and installation processes all come into play when you move legacy servers to newer hardware.
Hardware supportability for legacy servers
The first issue you’re likely to run into is that the virtual infrastructure may not support the operating system on your legacy servers. Microsoft does not support DOS or Windows NT Server in either physical or virtual environments, for instance. You should use a supported configuration for your legacy OS whenever possible, and even physical hardware has lacked support for many legacy OSes for years.
If you are planning a virtual server migration for a machine that’s running a legacy OS such as Windows NT or DOS, then you can’t perform a typical physical-to-virtual (P2V) conversion. Instead, you have to perform the virtual server migration process manually.
The problem with device drivers
Normally, when you install an OS on a virtual machine (VM), your virtualization platform uses Microsoft’s Hyper-V Integration Service or VMware Tools to supply the OS with drivers so it can recognize the virtualization hardware.
But these enlightening tools, as they’re called, won’t work with a legacy OS. Your only option for hardware support in legacy servers is therefore to use emulated drivers. Emulated drivers force the VM to emulate older, widely supported hardware devices. Hyper-V can emulate an NE2000 network card, for instance, which works well with Windows NT, DOS, Windows 2000, etc.
Legacy servers require manual install
Depending on which OS a legacy server is running, you probably won’t be able to perform a full backup and restore of the server before the migration. Instead, you’ll have to manually install the OS to the new VM, manually install the applications, back up the application’s data and restore it to the new virtual server. If you attempt to fully back up and restore legacy servers, you risk overwriting hardware-specific configuration settings, likely rendering the VM unbootable. Again, however, it depends on your OS and the capabilities of your backup software.
Manually installing the legacy OS and applications for legacy servers can be a big challenge. The first challenge with this virtual server migration process is finding the installation media. Unfortunately, depending on the age of the OS and the application, they may reside on floppy disks or on non-bootable CDs.
The legacy OS Windows NT Server, for example, shipped on non-bootable CDs. You had to launch the boot process from three floppy disks. If you lost or damaged the floppies, you could still copy the Windows NT installation files to the server’s hard drive and run the WINNT /B command. That, however, required you to boot the server to a DOS environment (and load DOS-compatible CD-ROM drivers).
The major virtualization platforms do support floppy drives and USB floppy drives, but you may have to jump through a few hoops to get the installation to work. Hyper-V and VMware, for example, won’t allow a VM to use a floppy drive directly. Instead, you have to use a virtual floppy disk file.
As you can see, there are many hurdles to migrating legacy servers to a virtual infrastructure. To avoid problems with virtual server migration, test the VM deployment in a virtual testing environment. That way, you get a feel for the virtual server migration process and can resolve any issues you encounter with the legacy OS.
Virtual server migration guide
Virtual machine migration FAQ: Live migration, P2V and more
What are best practices for virtual server migration?