Greenplum Database is architected to have no single point of failure. Internally the system utilizes log shipping and segment-level replication to achieve redundancy, and provides automated failover and fully online “self-healing” resynchronization. The system provides multiple levels of redundancy and integrity checking. At the lowest level, Greenplum Database utilizes RAID-0+1 or RAID-5 storage to detect and mask disk failures. At the system level, it continuously replicates all segment and master data to other nodes within the system to ensure that the loss of a machine will not impact the overall database availability. The database also utilizes redundant network interfaces on.
