Lesson 6 — Technology Evaluation, Business Continuity, Backup, and Recovery
Evaluate database technology by fit
Do not assume one database architecture fits every workload. Evaluate together:
- architecture and distribution;
- expected volume/velocity;
- transaction vs analytical workload;
- required function and platform compatibility;
- administration tooling;
- performance, scalability, resiliency, and security;
- support model and staff skills;
- vendor viability/reputation/support;
- licensing and total cost;
- organizational risk appetite.
A product that scales horizontally but lacks transaction commit/rollback is a poor fit for a point-of-sale process that requires transaction integrity. A pilot/POC should test critical requirements, not merely prove the software installs.
After selection, Technology Support manages installation, configuration, patches/upgrades, administration, training, Level 2 support, monitoring, and platform lifecycle. Track versions, licenses, support status, cost, and obsolete/unsupported technology as information assets.
Continuity is a proven capability
Every database/server should have a recovery plan for plausible loss/corruption scenarios. Business criticality determines restoration priority.
Backup choices
- backup frequency should reflect SLA/service expectations and data-loss risk/cost;
- full backups capture a complete backup set;
- incremental backups capture changes since a prior backup point;
- hot backup occurs while applications/database are running;
- cold backup occurs while the database is offline;
- protected copies should be separated from live database files and stored off-site where required.
A backup only on the same failed filesystem/storage array can share the primary system's failure mode.
The decisive rule: test recovery
A successful backup job proves a copy was created, not that the service can be restored. Recovery testing on non-production/failover infrastructure proves:
- backup files are readable;
- procedures are complete;
- credentials/configuration are available;
- staff can execute the process;
- restoration can meet the business need.
Memory chain: Criticality → SLA/backup → separate/off-site copy → recovery test → recovery/failover.
Source: pp. 188–193.