CompTIA DataSys+ (DS0-001)Business ContinuityHard
A database administrator is configuring a database for high availability using a cluster with shared storage. To prevent data corruption during a split-brain scenario where both nodes attempt to write to the shared storage simultaneously, which mechanism is essential?
- ALoad balancing
- BQuorum and fencing (STONITH)
- CNetwork segmentation
- DDatabase-level locking
Show answer & explanationAnswer & explanation
Correct answer: B. Quorum and fencing (STONITH)
Quorum mechanisms ensure that only a single active component of a cluster can operate, preventing split-brain. Fencing (also known as STONITH - Shoot The Other Node In The Head) is a critical component of quorum that forcefully removes a faulty or unresponsive node from the cluster's shared resources, preventing it from causing data corruption by writing to shared storage simultaneously with the active node.
Why the other options are wrong
- A. Load balancing distributes client requests and does not address the issue of preventing conflicting writes to shared storage during a cluster communication failure.
- C. Network segmentation isolates network traffic but does not directly prevent two nodes from attempting to become primary and write to shared storage if they lose communication with each other.
- D. Database-level locking prevents concurrent access to specific data records, not simultaneous write access to shared storage by conflicting cluster nodes.
Fencing (STONITH)
A high-availability mechanism that ensures a failed or unresponsive cluster node is truly isolated from shared resources to prevent data corruption (split-brain).
- Prevents split-brain scenarios.
- Forcefully disconnects/powers off a problematic node.
- Often implemented via power switches, storage locks, or network switches.
Memory trick: Fencing 'Fences' off bad nodes.