Will MWS ever support grid computing?

I wonder whether MWS will support grid computing in the near future. Many FDTD softwares support such a feature. For instance RSOFT's fullwave.The simulation domain is divided into smaller sub-domains. And each sub-domain is simulated on a seperate computer. The communication between the boundaries of each sub-domain is done through simple TCP/IP. This is a great future. It gives one the ability to simulate very large structures without the cost of simulation time. And also the simulation of very large structures, which can not be done on a single machine due to memory limitations, become possible.

But on the other hand, Fullwave has a drawback: the uniform rectangular mesh. In my case, this causes a great problem. The structure consists of electrically small features. But the overall structure is several wavelengths large. A simple comparison shows why the uniform rectangular grid is not suitable for such structures. One unit cell of such structure requires like 900 MB memory in Fullwave. On the other hand, MWS only requires around 150 MB of memory for the same structure. (These numbers are obtained by trying to get a similar accuracy). Lets say I have a cubical made of such unit cells. If I have 5 unit cells along z, 10 unit cells along y, and 30 unit cells along x. Approximately, I would need 1.35 TB memory If i were to simulate this structure by using fullwave. On the other hand MWS would require only 225 GB of memory. With some tweaking, I beleive the memory requirement for MWS could be reduced down to 100 GB. So If MWS supported grid computing, I would only need around 50 computers each with 3 GB of memory. On the other hand, doing such a computation with Fullwave is not feasible. Unless you want to build a supercomputer, of course. Well, I think I made my point