The problem is dealing with the application side.

Just one feature that's massive - how many systems have automatic import/export using Excel file formats. Converting those processes will be a huge undertaking themselves, let alone how many other things that will require re-engineering. The scope and scale of this is staggering.

A better effort would be to convert a single, small organization in government, then the scope is limited, but you get to build the fundamentals, and gain the experience of interfacing with extant systems.