| 摘要 |
A cluster computer system and method for distributed data storage enables distributed, reliable, low overhead mass storage systems. In particular, the system and method contemplate a plurality of computers connected in a network, each computer comprising a node having mass storage devices providing storage areas for data and redundancy information. Computer nodes access their locally attached mass storage devices independently and asynchronously, with a minimum of network transactions and storage overhead. The information stored on the mass storage devices is highly reliable due to the distribution of redundant information within the computer cluster. Redundancy information is used to reconstruct data associated with storage area access failures.
|
