1. Core Architecture Components
The power of RAC lies in its ability to present multiple servers as a single database to the end-user.
| Component | Function |
| Grid Infrastructure | The foundation layer that includes Oracle Clusterware and ASM (Automatic Storage Management). It manages the "membership" of nodes. |
| Interconnect | A dedicated, high-speed private network used for Cache Fusion. It allows nodes to ship data blocks to each other without hitting the disk. |
| Shared Storage | A SAN or NAS where the actual data files, redo logs, and control files reside, accessible by all nodes simultaneously. |
| SCAN (Single Client Access Name) | A virtual name that provides a single point of entry for clients, regardless of how many nodes are in the cluster. |
2. Global Resource Management
To prevent data corruption when multiple servers try to update the same record, RAC uses two primary background services:
Global Cache Service (GCS): Tracks the location and status of data blocks in the various instance caches.
Global Enqueue Service (GES): Manages "enqueues" (locks) to ensure transaction consistency across the cluster.
These services together maintain the Global Resource Directory (GRD), a "map" of which node holds which piece of data in its memory.
3. High Availability & Failover
One of RAC's biggest selling points is its ability to hide failures from the application.
Fast Application Notification (FAN): An "interrupt" mechanism where the database tells the application immediately when a node fails, so the app doesn't hang waiting for a TCP timeout.
Transparent Application Failover (TAF): A client-side feature that automatically reconnects a user session to a surviving node. If you were running a
SELECTquery when the node died, TAF can even resume the query on the new node.Application Continuity (AC): A more modern feature (12c+) that can replay uncommitted transactions (DML) after a failure, making the outage almost completely invisible to the user.
4. Key Benefits vs. Challenges
While RAC offers massive scale, it comes with increased complexity.
Benefits
Scalability: Add a server to the cluster to increase CPU/RAM capacity without taking the database offline.
Rolling Upgrades: Apply patches or OS updates to one node at a time while the others keep the business running.
Load Balancing: Automatically routes new users to the least-busy server in the cluster.
Scalability: Add a server to the cluster to increase CPU/RAM capacity without taking the database offline.
Rolling Upgrades: Apply patches or OS updates to one node at a time while the others keep the business running.
Load Balancing: Automatically routes new users to the least-busy server in the cluster.
Challenges
Cost: RAC is an "Extra Cost Option" on top of the Enterprise Edition license.
Complexity: Requires specialized networking (Interconnect) and storage (ASM/OCFS2) skills.
Application Design: If an application is not "RAC-aware" (e.g., heavy contention for the same data blocks), the overhead of moving blocks between nodes can actually make it slower than a single-instance database.
Cost: RAC is an "Extra Cost Option" on top of the Enterprise Edition license.
Complexity: Requires specialized networking (Interconnect) and storage (ASM/OCFS2) skills.
Application Design: If an application is not "RAC-aware" (e.g., heavy contention for the same data blocks), the overhead of moving blocks between nodes can actually make it slower than a single-instance database.
