4.3.2. High Availability and Testing

HA Approaches:

• Active-Active: Both systems handle traffic simultaneously (load balanced)
• Active-Passive: Standby sits idle until primary fails
• Clustering: Multiple servers act as one logical system; surviving nodes keep the service running if one fails
• Load balancing: Spreads requests across servers (see load balancer, 2.2.1); health checks let it route around a failed server

Fault tolerance means redundancy absorbs a failure with no outage—duplicate ISPs, power supplies, or devices remove single points of failure. In an active-passive firewall pair, stateful failover synchronizes the connection table to the standby so existing sessions survive failover; without it, sessions drop and must reconnect.

Testing:

• Tabletop exercises: Discussion-based walk-through of scenarios ("What would we do if...")
• Validation tests: Actually test failover procedures (fail over to DR site, verify operations)
• Full-scale drills: Simulate disaster conditions with actual recovery execution

Backup Types:

Incrementals are fastest and smallest to create but slowest to restore; differentials restore faster at the cost of growing each day. 3-2-1 rule: three copies of data, on two types of media, one off-site. A backup is only proven once it has been restored and verified.