3.4 Resilience and recovery explained

Availability is one third of the CIA triad, and objective 3.4 covers how architecture protects it. The exam expects you to choose the right resilience option for a stated requirement and cost.

High availability keeps a service running through the failure of a component. Load balancing spreads traffic across several servers, and clustering lets several servers act as one so a standby can take over. In an active/active setup all members serve traffic at once; in active/passive one waits idle until it is needed.

Site considerations cover where you recover. A hot site is fully equipped and current, so failover is fast, but it is the most expensive. A warm site has hardware and connectivity but needs recent data restored before it can take over. A cold site is space and power with little else, cheap but slow. Geographic dispersion puts sites far enough apart that one regional disaster can't hit both.

Platform diversity means not relying on one operating system or vendor, so a single flaw can't take everything down. Multi-cloud does the same across cloud providers. Continuity of operations is the plan for keeping essential functions going during a disruption. Capacity planning covers people, technology and infrastructure: enough trained staff, enough compute and enough network to cope with demand or with losing part of it.

Testing proves the plan works. A tabletop exercise is a discussion of a scenario with no systems touched. Failover testing actually switches to the backup. A simulation acts out an event. A parallel processing test runs the recovery site alongside production to check that its results match, without stopping production.

Backups get detailed attention. Onsite backups restore fast; offsite backups survive a disaster at the main site. Frequency decides how much data you can lose, which ties to the recovery point objective. Backups should be encrypted. Snapshots capture a point-in-time state quickly. Recovery has to be tested, not assumed. Replication copies data continuously to another location, and journaling records each change so you can replay to a point. A full backup copies everything; an incremental copies changes since the last backup of any kind; a differential copies changes since the last full backup.

Power is the last piece: generators for long outages, and an uninterruptible power supply to bridge the gap until the generator starts.