Oracle Real Application Cluster (RAC) is the successor to Oracle Parallel Server (OPS) and allows multiple instances to access the same database (storage) accessing and modifying the same data at the same time.
RAC provides fault tolerance, load balancing, and performance benefits by allowing the system to scale out, and at the same time since all nodes access the same database, the failure of one instance will not cause the loss of access to the database.
At the heart of Oracle10g RAC is the OCR and Voting Disk. All the nodes in the cluster must be able to access all of the data, redo log files, control files and spfile. At each instance level exist redo log file(s) and UNDO tablespace. All the instances in the RAC cluster must be able to access the controlfiles, redo logs and UNDO tablespaces to recover a node in the event of a system failure.
As of Oracle Database 10g RAC, data files, redo log files, control files, and archivelog files can reside on shared storage. Shared storage can be on raw-disk devices, NAS, ASM, or a clustered file system. At a minimum, a 1GB bandwidth for Oracle Interconnect and public network is required to successfully implement RAC. Typically, the networks cards (NICs) are aggregated to provide 2-4GBit bandwidth. Likewise, the network bandwidth for the public network is often aggregated to provide 2-4GBit bandwidth. With technologies such as Infiniband, the bandwidth can be amplified upto 10Gbit per second.
We recommend having at a minimum: 3 voting disks and mirroring the OCR. The voting disks must have at least 50% survivability for the RAC cluster to stay up. For example, if you have two voting disks and lose one of the two voting disks, you will only have 50% of the voting disks. In this particular case, the entire RAC cluster will crash.
Posted by: Charles Kim @ DBAExpert.com