Skip to content

Integrated Capstone — High-Volume Service Remediation

Situation

A company launches a high-volume service with:

  • one Production database;
  • a scaled-down Test environment;
  • daily backups stored on the same filesystem as the database;
  • no restore tests;
  • month-end load spikes;
  • customer Production data copied into sandboxes;
  • a plan to keep every record forever.

Leadership asks for a Chapter 6 remediation sequence.

Strong sequence

  1. Classify business criticality and define service/continuity expectations and restoration priorities.
  2. Separate backup storage, set SLA-backed frequency, secure appropriate off-site copies, and test recovery.
  3. Make Test sufficiently Production-like for the tests being performed; do not use Production as the test environment.
  4. Review Sandbox isolation and protected-data controls; mask/restrict Production-derived customer data or use synthetic/purpose-built data.
  5. Model capacity and growth including indexes, logs, mirrors/redundancy, and the storage impact of retention.
  6. Analyze usage patterns and performance bottlenecks; tune from evidence before structural redesign/denormalization.
  7. Define deliberate retention → archive → purge policies rather than indefinite retention by default.
  8. Establish monitoring/metrics, configuration control, and scripted change promotion.

Why this order matters

The weakest response is a technology-only redesign. Chapter 6 treats Data Storage and Operations as a controlled service: business continuity and requirements drive technology, environments, backup/recovery, capacity, performance, lifecycle, data protection, and governance.

Changed-fact test

For each step, state one changed fact that would alter the response. Examples: - off-site restores already tested successfully → focus shifts to another continuity weakness; - Test is already Production-equivalent → focus shifts from infrastructure similarity to test scope/data; - records have indefinite legal-retention requirement → “purge after N years” is no longer appropriate; - slowdown correlates to stale statistics after volatility → optimizer/statistics triage becomes immediate.

Source: Chapter 6, pp. 166–208.

← Scenarios 10–18 · Question Bank →