In grid storage or storage grids, multiple storage nodes are interconnected in such a way that each storage node can communicate with every other node without the data passing through a central switch. Several storage nodes share a common switch. In the grid storage concept, data resources are stored independently of platform and location.
Each storage node has its own storage media and processors to control them. With this storage virtualization, data can be divided into classes and stored according to importance, expiration date or other criteria. The user does not know where his data is physically stored, but always has direct access via his own management level and virtual address.
Grid storage, which can be seen as the equivalent of grid computing, ensures that utilization, availability and storage capacity can be adjusted flexibly and dynamically. It also features high fault tolerance and redundancy. If a storage node fails or a transmission path between two nodes is interrupted, the network can establish a transmission path to a redundant storage node. Even with fluctuating loads, the network provides optimal load balancing. Access to a node is independent of the network infrastructure and the network protocol used. Such a concept provides the appropriate functions for mirroring, snapshot, backup and data recovery.