Large quantities of computing resources go unused every day; many PCs, for example, are idle much of the time. Grid computing aims to tap these resources in much the same way as we use an electric power grid. This would allow us to solve large, computationally expensive problems cheaply. However, on large networks such as the Internet, it is not possible for all users to know and trust one another. Further, since resource donors may not receive any compensation, they have little incentive to accept code without confirmation that it will be safe to run. For these reasons, the ConCert project seeks to provide a theoretically sound basis for grid computing in an untrusted environment. To show that such a goal can be feasible, this project provides an initial implementation, using x86 TAL (Typed Assembly Language), of an actual framework for acquiring, verifying, and running grid code. TAL provides the ability to verify type safety; later, ConCert hopes to deliver capability to prove other properties, such as bounds on resources used. For this initial framework, requirements include taking full advantage of the fact that it will not be necessary to trust the code provider, tolerating failure in the face of unreliable links, and distributing code among machines differing in computing power.

32 pages

Return to: SCS Technical Report Collection
School of Computer Science homepage

This page maintained by reports@cs.cmu.edu