BigTable-like systems (HBase for instance) allows a table to be distributed transparently accross multiple servers. This works as follows:
- Each server perform the read/writes operations on distinct partitions (shards) of the same table.
- The DBA does not need to explicitely specify the server holding the partition in his query.
- When the DBA run his query, the API will route the request to the proper server.
- (The main benefit is scale-out scalability for write operations)
My assumption is that Oracle can not do so, but I am not an Oracle expert. References to infirm or confirm this hypothesis are much appreciated.