Lesson 9 — Governance, Metrics, and High-Value Decision Rules
Four metric families
| Family | What it answers | Examples |
|---|---|---|
| Storage | What footprint/workload are we carrying? | database count, storage use, containers, pages/blocks, queues, service usage |
| Performance | How fast/how much work? | transaction frequency/quantity, query response, API/service performance |
| Operational | Are operating processes healthy? | retrieval time, backup size/success, DQ measures, availability, recovery outcomes |
| Service | How well is support working? | incidents/issues, resolution time, escalations, recurring support problems |
Information-asset tracking covers licenses, support costs, leases/fixed cost, TCO, and obsolete/unsupported/expensive technologies.
Data Audit vs Data Validation
- Data Audit: evaluate a dataset/process against defined contractual, methodological, compliance, or similar criteria.
- Data Validation: evaluate stored data against agreed acceptance criteria for quality and usability.
Contract requirement → audit. Fitness/quality acceptance → validation.
24 high-value decision rules
- Technology Support vs Operations Support = platform vs stored-data service/process.
- Database vs Instance vs Schema vs Node = collection / running software / object grouping / distributed computer.
- Best practice vs rigid rule = strong default with justified exceptions vs blind project blocker.
- Production DBA vs Application DBA = production reliability vs application DBs across environments.
- Procedural vs Development DBA = DBMS procedural code vs design/sandbox/development focus.
- Centralized vs Distributed = one location/system vs multiple nodes/systems.
- Federated vs non-federated distributed = autonomous components vs centrally controlled components.
- Federation vs Replication = unified access without required duplication vs maintained copies.
- VM DB vs DaaS vs managed hosting = who actually runs/maintains the database service.
- ACID vs BASE vs CAP = transaction guarantees vs eventual-consistency posture vs partition trade-off.
- Row vs Column = whole-row transaction access vs selected-column analytical scans.
- Dev/Test/Sandbox vs Production = safe change/validation/experimentation vs live business processing.
- Archive vs Purge = retained/retrievable vs irreversible removal.
- Retention vs Backup = how long data must remain vs recovery copy/frequency.
- CDC vs Replication = detect/move deltas vs maintain copies.
- Sharding vs Replication = partition into independent chunks vs duplicate same data.
- Resiliency vs Recovery = continue despite failure vs restore after failure.
- Backup success vs Recovery proof = copy created vs actual restoration tested.
- Physical vs Functional config audit = installed as designed vs functions/performs as required.
- Index improvement vs over-indexing = read benefit vs write/storage maintenance cost.
- Test realism vs protection = realistic patterns do not override masking/security duties.
- Migration technical completion vs verified migration = load ran vs correctness/completeness reconciled.
- Data Audit vs Validation = compliance/method criteria vs quality/usability acceptance criteria.
- Storage vs Performance vs Operational vs Service metrics = footprint vs speed/throughput vs process health vs support responsiveness.
Final teach-it-back
Explain Chapter 6 as one system: continuity demands available/intact/performant stored data; Technology Support chooses and operates the platform; Operations Support controls data/environments through change, backup, recovery, retention, copies, testing, performance, and migration; governance makes responsibility explicit; metrics prove whether the service is healthy.
Source: pp. 206–208 and chapter-wide distinctions.