Exchange nodes have been down before, but this the worse I've ever seen. The ability to keep nodes running should be getting better and not worse. There is a fundamental issue that needs to be addressed.