| What Is Oracle Real Application Clusters 10G? | | | | unused IP address on the same subnet as LAN. This |
| Oracle Real Application Clusters is an option of Oracle | | | | address is used by application to connect to the RAC |
| Database that was first introduced with Oracle | | | | database. If a node fails, the virtual IP is failed over |
| 9i.Oracle Real Application Clusters is now proven | | | | to another node in the cluster to provide an |
| technology used by thousands of customers in every | | | | immediate note down response to connection |
| industry in every type of application. Oracle RAC | | | | requests. |
| provides option for scaling applications beyond the | | | | Benefits of Real Applications Clusters |
| capabilities of a single server. This allows customers | | | | Scalability |
| to take advantage of lower cost commodity | | | | Scalability is the ability to add additional nodes to Real |
| hardware to reduce their total cost of ownership and | | | | Application Clusters and achieve markedly improved |
| provide a scalable computing environment that | | | | performance. Real Application Clusters can take |
| supports their application workload. Oracle RAC | | | | advantage of additional equipment and harness the |
| provides the ability to remove the server as a single | | | | processing power of multiple systems. |
| point of failure in any database application | | | | All servers in the cluster must run the same operating |
| environment. | | | | system and same version of oracle but they do not |
| Real Application Clusters Architecture | | | | have to be exactly the same capacity. |
| A RAC database is a clustered database. A cluster is | | | | High Availability |
| a group of independent servers that cooperate as a | | | | The term high availability refers to systems with |
| single system .Clusters provide improved fault | | | | redundant components that provide consistent, |
| resilience and modular incremental system growth | | | | uninterrupted service, even in the event of hardware |
| over single symmetric multiprocess(SMP) systems. In | | | | or software failures. In most high availability |
| the event of a system failure,clustering ensures high | | | | configurations, nodes are isolated from each other so |
| availability to users. Access to mission critical data is | | | | a failure at one node does not affect the entire |
| not lost. | | | | system. In such a case, surviving nodes compensate |
| Redundant hardware components such as additional | | | | for the loss of the failed node through recovery |
| nodes,interconnects and disks allow the cluster to | | | | andthe system continues to provide data access to |
| provide high availability. Such redundant hardware | | | | users. This means data is consistentlyavailable, more |
| architectures avoid single points-of-failure and provide | | | | so than it would be with a single node upon node |
| exceptional fault resilience. | | | | failure. High availability also implies increased database |
| With Real Application Clusters,we de-couple the | | | | availability. |
| Oracle Instance (the processes and memory | | | | Transparency |
| structures running on a server to allow access to the | | | | The concept of transparency is the functional |
| data) from the Oracle database(the physical | | | | equivalent of single instance Oracle and shared |
| structures residing on storage which actually hold the | | | | configurations that use Real Application Clusters. |
| data,commonly known as data files). | | | | Applications that run on single instance Oracle |
| A clustered database is a single database that can be | | | | execute with the same results using Real Application |
| accessed by multiple instances. Each instance runs on | | | | Clusters. An Oracle database can be configured to |
| a separate server in the cluster. When additional | | | | execute in three different modes: |
| resources are required, additional nodes and instances | | | | • Single instance exclusive |
| can be easily added to the cluster with no downtime. | | | | • Shared with a single instance |
| Once the new instance is started, application using | | | | • Shared with two or more instances |
| services can immediately take advantages of it with | | | | Buffer Cache Management |
| no changes to the application or application server. | | | | Within a single instance, Oracle stores resources, such |
| Oracle Clusterware | | | | as data block information, in a buffer cache that |
| Oracle provides Oracle Clusterware, a portable | | | | resides in memory. Storing this information locally |
| clusterware solution that is integrated and designed | | | | reduces the disk Input/Output (I/O) necessary for |
| specifically for Oracle Database. You no longer have | | | | database operations. Since each node in Real |
| to purchase third party clusterware in order to have | | | | Application Clusters has its own memory that is not |
| a RAC database. Oracle Clusterware monitors and | | | | shared with other nodes, Real Application Clusters |
| manages Real Application Cluster databases. When a | | | | must coordinate the buffer caches of different |
| node in the cluster is started, all instances,listeners | | | | nodes while minimizing additional disk I/O that could |
| and services are automatically started. If an instance | | | | reduce performance. The Oracle Global Cache Service |
| fails the clusterware will automatically restart the | | | | technology maintains the high-performance features |
| instance so the service is often restored before the | | | | of Oracle while coordinating multiple buffer caches. |
| administer notices it was down. | | | | Fast Commits, Group Commits, and Deferred Writes |
| Hardware Architecture | | | | Fast commits, group commits, and deferred writes |
| Oracle Real Application Clusters is shared everything | | | | operate on each instance in Oracle and work the |
| architecture. All servers in the cluster must share all | | | | same whether in exclusive or shared mode. Oracle |
| storage used for a RAC database. The type of disk | | | | only reads data blocks from disk if they are not |
| storage used can be network attached | | | | already in the buffer caches of one of the instances. |
| storage(NAS),storage area network(SAN), or SCSI | | | | Because data block writes are deferred, they often |
| disk. Your storage choice is dictated by the server | | | | contain modifications from multiple transactions. |
| hardware choice and what your hardware vendor | | | | Optimally, Oracle writes modified data blocks to disk |
| supports. Choosing a storage system that will provide | | | | only when necessary: |
| scalable I/O for your application an I/O system that | | | | • When the blocks have not been used recently |
| will scale as additional servers are added to the | | | | and new data requires buffer cache space (in shared |
| cluster. | | | | or exclusive mode) |
| A Cluster requires an additional network to the Local | | | | • During checkpoints (in shared or exclusive mount |
| Area Network (LAN) that a database server is | | | | mode) |
| attached to for application connections. A cluster | | | | • When another instance needs the blocks (only in |
| require a second server is attached to for application | | | | shared mode) |
| connections. A cluster requires a second private | | | | • Cache copies of dirty blocks (changed blocks) |
| network commonly known as the interconnect. | | | | across the interconnect (write/write consistency). |
| Oracle recommends that you use 2 network | | | | Row Locking and Multi version Read Consistency |
| interfaces for this network for high availability | | | | Oracle’s row locking feature allows multiple |
| purposes. A network interface bonding external to | | | | transactions from separate nodes to lock and update |
| Oracle should be used to provide fail over and load | | | | different rows of the same data block. This is done |
| balancing. The interconnect is used by cluster for | | | | without any of the transactions waiting for the |
| inter-node messaging. The interconnect is also used | | | | others to commit. If a row has been modified but |
| by RAC to implement the cache fusion technology. | | | | not yet committed, then the original row values are |
| Virtual Internet Protocol Address (VIP) | | | | available to all instances for read access. This is called |
| Oracle RAC 10g requires a virtual IP address for each | | | | multi version read consistency. |
| server in the cluster. The virtual IP address is an | | | | |