The material uncovered above has some very clearly worded & comprehensive discussion on GRID & also its relationship (& dependance) on Web Services.

Looking at the 'Heterogenous' aspect of GRID the aspect that gets really interesting is software technologies that depend on Virtual Machines (i.e. Java & now Microsoft's VM).

In living up to the goals of GRID as clearly stated in most of the base docs, software that runs on a VM can easily be made to run anywhere on a GRID.

The fact that IBM Microsoft & Sun all have mature & growingly used VMs suggests to me that GRID can be made to work and work well. The Web Svcs part just solves the issue of message passing & standard interface techniques. If every PC & Computer can advertise its' resources & current load, as a simple Web Service, then the management across heterogenous computers, becomes very easy if all or parts of jobs can me moved & run on otherwise incompatible hardware technologies by taking advantage of common VMs.

Seems to me the core elements of GRID are :-

1) Lots of computers
2) Ever faster communications links
3) XML as the means of passing self describing data (without the overheads of shared objects space)
4) Web Services (building on XML) to provide a technology agnostic and standard rpc mechanism
5) Vitual Machines as a layer above the underlying chip technologies & architectures (Cisc, risc, parallel, SMP, Clustered etc:)
6) A desire to leverage the massive unutilised horsepower of 'all those computers' without worrying about boundaries & incompatible technologies

WILL IT SUCCEED.
I can see this succeeding because unlike DCE & CORBA etc: - everyone who is important & influential in IT is more or less adopting the same core technologies.

Also and very importantly, implementing XML/Web Services & GRID, in an era of business globalization, will drive big now opportunities & spending (what keeps IT vendors alive).

Cheers

Doug Marker