We chose the word ‘enclave’ to represent this notion of a group of nodes that can be collectively managed. We use a different word because we don’t want to overload the word partition or job because that immediately has a lot of connotations that people follow behind it with. So instead we just said let’s assume that for every large group of nodes we assume that there’s some management there – some place where I can run some meta code. And we’ll call that large group an enclave. And the beauty of computer science is that we can do this recursively. We can just start at the top of the machine and say the entire machine is one enclave – which is the system console – and then you can break it up and for a particular job you might have another enclave and that job might have both a climate modeling piece and an economic modeling piece that might run on separate partitions.
Pete Beckman Interview on the Argo OS for Exascale
interview from The Exascale Report, Pete Beckman from Argonne National Lab discusses the Argo Exascale Operating System.In this