ORACLE RAC principle in an application environment, all servers use and manage the same database in order to spread the workload of each server, requiring at least two more servers on the hardware, and a shared storage device. It also requires two types of software, one is cluster software, the other is the RAC component in the Oracle database. At the same time all the servers on the OS should be the same type of OS, according to the load-balanced configuration strategy, when a client sends a request to a service listener, the server according to our load balancing strategy, will send the request to the local RAC component processing may also be sent to another server's RAC component processing, after processing the request, the RAC will use the cluster software to access our shared storage device.
Logically, each node participating in the cluster has a separate instance that accesses the same database. Nodes communicate through the communication layer (communication layer) of the cluster software. At the same time, in order to reduce IO consumption, there is a global caching service, so the instance of each database retains a same database cache.
The characteristics of RAC are:
Each node's instance has its own SGA.
Each node's instance has its own background process.
Each node of the instance has its own redo logs
The instance of each node has its own undo table space
All nodes share a copy of Datafiles and Controlfiles
Oracle has also proposed a caching fusion technology (cache fusing)
There are two purposes.
1. Ensure the consistency of the cache
2. Reduce the consumption of shared disk IO
As a result, multiple nodes in a RAC environment retain the same DB CACHE
The principle of Cache fusion:
****************************************
1. One of the nodes reads a block into the DB cache from the shared database
2. This node will cross DB block copy at all nodes
3. When any one of the node cache is modified, it will be cached between the nodes.
4. Results of consistent final modifications to the storage will also be written to disk
Clusterware components
*******************
There are four kinds of service
CRSD-Cluster Resource services
CSSD-Cluster Sync Service
EVMD-Event Management Services
OPROCD-node detection monitoring
There are three types of resource
VIP-Virtual IP address (fictitious IP)
Ocr-oracle Cluster Registry (cluster registration file), recording information about each node
Voting disk-establishes Quorum (voting disk), the arbitration mechanism is used to arbitrate the behavior that multiple nodes write concurrently to a shared node, in order to avoid conflicts.
‑
Components of RAC
************
Additional processes are provided to maintain the database
Lms-gobal Cache Service Process global caching services processes
Lmd-global Enqueue Service Daemon global Query Services daemon
Lmon-global Enqueue Service Monitor Global query services monitoring process
Lck0-instance Enqueue Process Instance query processes