Some Basic RAC Concepts

I found previously I have developed some reading notes about RAC, I am reviewing them today, becuase I will have a technical phone screen tonight which is said about Oracle DBA with RAC experiences. Hopefully review those notes will refresh my mind.

Below are part of the notes about some basic RAC concepts:

13. RAC concepts

13.1 What is RAC ?
RAC architecture consisted of:
(1) Clustered hardware:
- Multiple instances accessing the same database
- One instance per node
- Interconnected nodes act as a single server

(2) Real Application Clusters software:
- Manage coordinated data access and data changes among instances
- Each instance sees a consistent image of the database
- Pass coordination information and data images among each interconnected nodes.

13.2 Why use RAC
- High availability
- Scalability

13.3 RAC specific processes
At RAC instance level
* LMON – GES Monitor
* LMD0 – GES Daemon
* LMSx – GCS processes: manages resources and provides inter-instance resource control.
* LCK0: lock process
* DIAG: Diagnosability process
At cluster level:
* CRSD and RACGIMON: are engines for high-availability operations
* OCSSD: Provides access to node membership and group services
* EVMD: Scan callout directory and invokes callouts in reactions to detected events
* OPRROCD: Is a process monitor for the cluster: (not in Windows and Linux)

13.4 Global Resource Coordination

- Global Resource Directory (GRD):
* Each instance maintains a part of GRD
* Each particular resource has a master (the instance that manages all the information about this resource)
- Global Cache Service (GCS)
* buffer cache
* cache coherency – Cache Fusion algorithm
- Global Enqueue Service (GED)
* dictionary cache locks
* library cache locks
* deadlock

13.5 Object Affinity and Dynamic Remastering

- GCS enables the database to automatically adapt and migrate resources in the GRD
- GCS keeps track of the number of GCS requests on a per-instance and per-object basis.
- Master a buffer cache resource on the instance where it is mostly accesse

13.6 RAC database files
- Must be on local storage: Oracle cluster software (ORA_CRS_HOME)
- Can be on shared but better on local storage: Oracle RDBMS and ASM software
- Must be on shared storage (e.g. ASM disk groups, CFS )

RAW devices:
* OCR files - on raw devices
* Voting files (disks) – on raw devices

ASM Disk group: +DATADG
* Data files
* Temp files
* Control files
* SPFILE
* Undo tablespace files ( each instance has its own dedicated ones)
* Redo log files ( at least two groups for each instance)

ASM DISK group: +FLADG
* Flash recovery area files
* Change tracking file
* Archived log files* Control file (duplicate