Skip to content
Ameen Abozaid
ARC-031

Resilience architecture

  • 5 Lessons
  • 3 h practice (estimate)
On this page 9 topics

Overview

Topics

  • Failure scenarios
  • fault isolation
  • continuity plans

Learn first

3 h of practice (estimate)

Learning outcomes

  1. Enumerate failure scenarios for a whole solution, including third-party and on-premises dependencies.
  2. Isolate faults so a failing component degrades a feature rather than the business.
  3. Write business continuity plans for processes when systems are unavailable (e.g., stores selling offline).
  4. Define resilience tests that prove the design.

Lessons

  1. 01Resilience beyond infrastructure
  2. 02Failure scenarios for enterprise solutions
  3. 03Fault isolation and degradation
  4. 04Business continuity planning
  5. 05Demo: keeping 120 stores selling when the central platform is down

Lab and exercises

Lab

ARC-031-LAB1 · Larkspur resilience

Outcomes: 1–4 · Time: ~3 h

  1. List Larkspur's critical business processes and map their dependencies, including third-party and on-premises ones.
  2. Build the failure-scenario catalog with impact, detection, and response.
  3. Design degradation for offline store sales, payment-provider failover, and ERP outage queueing.
  4. Write continuity plans per critical process.
  5. Define resilience tests that prove each degraded mode.

Acceptance: every critical business process has a documented degraded mode.

Exercises (case-based; six exercises)

IDOutcomeTask
ARC-031-EX011Larkspur's web checkout depends on six services (basket, pricing, stock, payment provider, ERP, tax). Enumerate ten failure scenarios across them, including slow and wrong-answer cases.
ARC-031-EX021The ERP is down every Sunday 02:00-04:00 for maintenance and sometimes overruns. Add this to the scenario catalog with impact and response.
ARC-031-EX032The tax service times out for 30 seconds under load and the checkout threads pile up. Design the isolation.
ARC-031-EX042Define the degraded modes for ERP unavailable, recommendations unavailable, and stock service unavailable, and what the customer sees.
ARC-031-EX053Write a one-page continuity plan for stores when the central platform is unreachable for four hours.
ARC-031-EX064Design three resilience tests for the payment-provider failover design: what you break, what you measure, pass criteria.
Show solutions
  • EX01 Examples: payment provider down; payment provider slow; tax service wrong rates; stock service stale; ERP unreachable; basket cache loss; bad deployment of pricing; database failover; duplicate callback; region loss. Each with impact and detection.
  • EX02 Planned maintenance window; orders queued durably; stock shown from platform's last-known view; postings replayed after; alert if the window overruns; capacity plan for the catch-up burst.
  • EX03 Timeout of a couple of seconds; circuit breaker; bulkhead so tax calls cannot take all threads; fallback to cached rates with flag for review; alert on open breaker.
  • EX04 ERP down: accept orders, queue postings, show "last known" stock. Recommendations down: hide the block. Stock down: show availability from cache with "confirm at checkout" or limit to in-stock-by-default items; message honest and short.
  • EX05 Trigger, authority to declare, offline selling steps and limits, manual receipts if needed, cash-only fallback, communication to stores, reconnection and reconciliation, owner and after-action review.
  • EX06 Block the primary provider endpoint: failover within target time, success rate recovers; inject elevated latency: breaker opens; restore primary: traffic returns without duplicate charges. Measure error rate, time to switch, duplicate payments.

Workplace evidence: a continuity plan or degraded mode you designed that was used or tested.

Quiz

Outcomes: O1 failure scenarios · O2 fault isolation and degradation · O3 business continuity · O4 resilience tests.

A self-check for you: nothing is graded, saved or sent anywhere. Choose an answer, then check it. You can try again as often as you like.

  1. 1Resilience for an enterprise solution should start from:O1
  2. 2The ERP has weekly maintenance windows. The catalog should:O1
  3. 3A slow dependency is often more dangerous than a dead one because:O1
  4. 4A bulkhead is:O2
  5. 5A degraded mode is:O2
  6. 6The ERP is down; the platform should preferably:O2
  7. 7A store continuity plan must include:O3
  8. 8Offline sales replayed after reconnection require:O3
  9. 9A resilience test for payment failover should:O4
  10. 10A critical process without a documented degraded mode:O4

Practical part

Scenario walkthrough with the assessor over the lab catalog and continuity plans plus exercises ARC-031-EX01 to EX06. Pass: quiz at least 8/10, at least five of six exercises correct, and a lab where every critical business process has a documented degraded mode.

Common mistakes

Resilience that covers servers but not processes; assuming the ERP is always available.

Assessment

Scenario walkthrough with the assessor.

Workplace evidence

A continuity plan or degraded mode you designed that was used or tested.

Career relevance

Resilience is a top concern of business stakeholders.

Type to search levels, phases, modules and lessons.