There is a misunderstanding of realtime computer architecture. Especially the layer of microkernels under the Linux or XP kernels and the layer of application above. Virtualization above the kernels have too much code overhead to be efficient.

When you have hard realtime microkernels, these packets commands the monolithical kernels of Linux and XP(BSD) kernels. So Xen and Virtualbox are applications, not communications of commands like real time packets.

You never want to use Xen or virtualbox to run Vista; you want to use WinCE microkernels to run Linux applications supported by Linux kernels.

Having said that, think of realtime packets as virtual private network in the enterprise system.

Novell is developing SleRT with CCUR, and the hard thread packets can run Vista applications.

Linux FireFox has soft realtime threads to use multicore cpu efficiently. But Linux still needs hard realtime packets and semi hard threads in hdd for synchronization of data uniquely for each thread on the fly.

Realtime architecture needs to be understood before Vista can be used efficiently with other monolithical software which has tremendous code overhead.

Once realtime microkernels set up hard(threads for cores), semi hard(threads for hdd virtual memory) and dynamic soft(longdistance connections on the net) threads, communications by realtime packets make every thing virtual and fastest possible. Loadbalancing becomes automatic via dynamic soft thread creation. Priority is just one additional thread for lower interrupt request level(IRQ). Most realtime operating system such as QNX etc. are soft realtime. CCUR has hard realtime experienced microkernels now for multicore.

Footnote:
KVM uses ethernet packets in the kernel to communicate with external devices and servers. Dotnet packets in browser communicate with websites. Both work in the dram ramdisks. Microkernel packets work in L2 cache in the cpu for load balancing in the threads of L2 cache protected memory via the extra pipeline either hypertransport or hyperthreading within an allocated slice of time(realtime definition). Then use the operating system kernels to communicate with internal devices in the computer.Then uses ethernet or wireless or broadband to communicate with external devices and websites.

In realtime operating system, the efficiency is dependent on the programmers' experience. Inexperienced programmers have to allow larger protected memory for future unforeseen larger packet sizes to be communicated without lossiness.

Realtime operating system runs simultaneous processing of data at the same time, multicores or multi-computers. Monolithical operating system runs multicores symmetrically. There is a clear advantage of data processing efficiency.

Since L1 cache is 32/64 kB, the thread/packet protected memory has a minimum size requirement. Of course, when we cheat to get more speed, we double the cache size and separate the input from the output cache. For longer packet sizes multiple threads for segmented data to be reassembled is necessary.