The main differences I see between “Grids” and the “Cloud” are the concepts of “virtual organizations” and “sharing of resources” that the Grid community has often used to drive its standardization efforts. While academic institutions seem to be more than happy to share their IT infrastructure with collaborators, the industry is mostly focusing on offering IT infrastructure or utility computing as a service, completely abstracting the use of the underlying hardware resources and the related management operations.
I remember a meeting in San Francisco where some key IBM OGSI/WSRF folks were trying to make the case of protocols for managing virtual resources remotely and Jim Gray’s question, which was along the lines: Why would anyone want to allow the remote control of hard drives across organization boundaries? Indeed, with cloud computing, we, the consumers of the utility resources, don’t care about the hard drives, the compute hardware, their management.
Those building the internal infrastructures to support Cloud services might care, of course. In a conversation with a major service provider’s IT infrastructure manager, it was explained to me how standardization of systems management software greatly helps in bringing the IT costs down, at least for those who cannot afford to build their own platforms, like Microsoft, Google, etc.