By linking with a wide range of cluster software and specifying the standby server in advance, HiRDB enables quick failover if a server failure occurs. Use of the HA monitor as the cluster software reduces the time required to detect a failure. By linking with HA Booster, HiRDB also realizes quick and stable failover in as little as several seconds*1 regardless of the number of disk resources for switchover. Furthermore, during the failover processing, the transaction queuing functionality receives new transactions, enabling the service to be continued without users being aware of the failure.
HiRDB uses a Shared Nothing architecture that can improve parallel processing capability in a linear manner by increasing the number of servers. HiRDB also supports Active-Active Cluster (for system switchover that requires no standby server) in which, if a failure occurs, the workload can be distributed to the remaining running servers. This reduces idle assets and enables both the high reliability and high availability. .
With HiRDB disaster recovery, the database is sent asynchronously to the remote site, so that only the system logs (update information) are transferred synchronously. This special "real-time SAN replication" functionality can minimize impact to online services running on the main site, for operation in which performance is barely degraded at all. HiRDB's own checkpoint methodology is also used to guarantee restart points on the remote site. This enables performance to be enabled with zero transaction defects, so that systems can be set up to guarantee valid transaction results.
See larger picture in a new window
You can copy the data to the remote site using the data updating functionality of the HiRDB Datareplicator software that enables copies between servers. This functionality transfers only the system log (update information), so that real-time data updating does not affect the online service performance. This functionality updates the data on a transaction basis to assure data integrity at the remote site. This functionality does not entail either the installation of special hardware or an increase in communication costs. You can implement the disaster recovery plan at a minimum cost.
