Exposes memory capacity available in remote locations as a pool of memory and shares it across multiple servers over the network. This architecture decouples the available compute and memory resources, enabling independent resource allocation in the rack cluster.