With Java, there exists something called Network Attached Memory (a Terracotta product). It basically increases the effective RAM of any single machine by allowing it to use RAM on another machine. But that's Java.
Is there an equivalent thing for C++? Specifically, I have a cluster of nodes with a moderate amount of memory (8 gbytes each) and a very fast network (10 microsecond small message latency and 750 mbytes/second large message bandwidth). I have a C++ application that needs tens of gbytes of RAM. Is there some kind of product, or library, that would allow the C++ application to seemlessly use the memory on other nodes as if it had a much larger RAM?
James