Despite the promise of vnode stacking, not one of several proposed implementations [Rosenthal90,Rosenthal92,Heidemann94,Skinner93] has made it into mainstream operating systems, even though several of the proposals were made by an operating system vendor (Sun Microsystems).
All previous proposals for vnode stacking required substantial changes to the definitions of the vnode and the vnode interface. These proposals did not meet with wide acceptance, for a few reasons:
Additionally, the vnode stacking proposals have always been linked to a particular operating system, and hence unportable. For more details see Section sec-bg-barriers-commercial.
Although these objections to the vnode stacking concept are not technical, they are fundamental. However, this thesis proposal will present another technical solution which avoids most of the above problems.