Geometry/mesh generation is not compute-intensive in the sense of requiring
large CPU time or even the fastest CPUs available. Rather, it is more an
assembly: a collection of relatively small operations on elements of a growing
object, itself an aggredation of smalled objects. And these smaller component
objects are reusable and editable to form still other final objects. And,
although completely hands-off operation is alays the ultimate ideal, reality
is that interactive steering - and some consultation among application users
and those more experienced in the generation process - will remain the case
for some time, because the geometry/mesh solution is not unique for any
application. Thus, geometry/mesh generation is, by its very nature, a
particularly good candidate for interactive distributed operation on both the
server and client side: multiple servers providing specialized resourses for
assembly and multiple clients participating in steering and consultation, with
a middle-tier locating and brokering resourses.

To be more specific, a library of NURBS for surface generation might reside on
a server at a center specializing in that aspect of the overall process and
maintaining such a library, while operations for the adaptive location of mesh
points on a surface might reside at another center with that specialty. Then
volume mesh operations might be at a third center, etc. Continuing, the
applications users doing computational science might be at one university
(or several universities), in consultation with mesh expertise at another
university.

Web-based portal architecture is especially well suited to this interactive
distributed operation, since the component operations and objects can be kept
and maintained close to the expertise required therefor rather than having to
be updated and broadcast to all potential users. This is particularly important
also to allow for the continual open source enhancement and expansion that isintended here.