VA wants to put all VistA instances on one [virtual probably] "node". This could facilitate a slow migration towards a heterogenous environment front-ended by some kind of master broker. On the call mention was made of the fact that different sites configure VistA to run differently. This is ameliorated by site parameter files, which allow for multiple sets of parameters by site, that would allow each user to be associated with a particular site on log on through a broker and view VistA through the lens of those parameters. In addition, access to non-VistA applications through a broker would allow blending of data from different applications databases. While this woud take time, there are many in government that are philosophically oriented towards purchase of goods and services from private industry and they will spend a lot of money in order to do this. Already, the core of VistA (FileMan and Kernel) are being released by private organizations paid by the VA and the only part of the work done by VA employees is something that can normally be thought of contractual, product validation / quality control.
I could easily make a plan to stage a migration of VistA/CPRS to such an environment, given enough money and, though the VistA database may survive relatively intact, it would be surounded by other systems (data vultures) feeding off its body (carcass). Commercial companies are very good at exploiting resources they did not create, often leaving behind their delitrous for the public to clean up. It gives them a commercial advantage.