Skip to main content

ATAM Scenarios

1. Performance - Responsive User Experience​

Scenario P1 - Peak-time product search​

  • Source: Customer
  • Stimulus: Submits a product search query with filters
  • Environment: Black Friday campaign traffic (5-10x normal load)
  • Artifact: Search and product listing path
  • Response: System returns ranked results and facet counts
  • Response Measure: Average response time 500-800ms; p95 ≤ 2s; error rate < 1%

Scenario P2 - Cart and checkout interaction speed​

  • Source: Customer
  • Stimulus: Adds a product variant to cart and proceeds one checkout step
  • Environment: Normal business hours load
  • Artifact: Cart service and checkout flow
  • Response: Cart updates and checkout step confirmation are shown
  • Response Measure: Average ≤ 1s for add-to-cart and checkout step APIs; p95 ≤ 2s

Scenario P3 - Performance regression after release​

  • Source: DevOps monitoring system
  • Stimulus: Detects sustained increase in checkout latency after deployment
  • Environment: Production, steady traffic
  • Artifact: Checkout API and dependent services
  • Response: Team identifies bottleneck and rolls back or mitigates before major business impact
  • Response Measure: Regression detected within 5 minutes; mitigation started within 15 minutes

2. Availability & Reliability - High System Availability During Business Hours​

Scenario A1 - Business-hours service outage​

  • Source: Infrastructure fault
  • Stimulus: One application instance fails unexpectedly
  • Environment: 10:00 on a normal weekday
  • Artifact: Customer-facing web and API entry points
  • Response: Traffic is routed to healthy instances with minimal user-visible impact
  • Response Measure: No full outage; user-visible error rate remains < 1%; service recovers within 5 minutes

Scenario A2 - Non-critical service failure during peak​

  • Source: Internal service dependency
  • Stimulus: Review service becomes unavailable
  • Environment: High-load promotional event
  • Artifact: Product detail page composition
  • Response: Page degrades gracefully (reviews hidden/unavailable) while shopping and checkout continue
  • Response Measure: Browse, cart, and checkout remain operational; no complete request failure caused by review outage

Scenario A3 - Failed deployment in business hours​

  • Source: Release pipeline
  • Stimulus: New version introduces elevated error rate after rollout starts
  • Environment: Business hours, live production
  • Artifact: Deployment and release control mechanisms
  • Response: Rollout is halted and traffic shifted to stable version
  • Response Measure: Detection within 5 minutes; rollback/traffic shift completed within 10 minutes; no prolonged outage

3. Data Integrity & Consistency - Reliable Order Processing​

Scenario D1 - Duplicate submit during checkout​

  • Source: Customer and browser/network behavior
  • Stimulus: Customer retries payment submission due to temporary UI delay
  • Environment: High latency or transient network issues
  • Artifact: Order creation and payment processing flow
  • Response: System prevents duplicate order creation and either reuses or safely resolves existing transaction state
  • Response Measure: Exactly one valid order recorded; no duplicate charge

Scenario D2 - Concurrent purchase of low-stock item​

  • Source: Multiple customers
  • Stimulus: Several checkout attempts for last available units
  • Environment: Peak traffic with high contention on inventory
  • Artifact: Inventory reservation and checkout commit path
  • Response: System enforces stock limits and rejects or backorders excess attempts safely
  • Response Measure: No oversell due to stale inventory in checkout path; inventory reflects committed sales within 30 seconds

Scenario D3 - Partial failure across order and payment​

warning

Does it make sense to evaluate this scenario in this environment?

  • Source: Payment provider timeout/failure
  • Stimulus: Payment confirmation fails after order workflow starts
  • Environment: Production incident conditions
  • Artifact: Transaction orchestration between order and payment
  • Response: System leaves no partial order state and executes compensating/rollback behavior
  • Response Measure: 100% of affected transactions end in either fully confirmed order or fully rolled back state

4. Scalability - Handling User Spikes and Business Growth​

Scenario SC1 - Flash sale traffic surge​

  • Source: Externally controlled marketing campaign (e.g. TV advertisement) draws in customers
  • Stimulus: Sudden, unexpected increase to 10x concurrent users within minutes
  • Environment: Ongoing promotional event
  • Artifact: Customer-facing browsing, cart, and checkout services
  • Response: Capacity scales up to maintain acceptable service
  • Response Measure: Core user journeys remain available; performance targets do not drop below p95 ≤ 2-3s, and normalize within 2 minutes

Scenario SC2 - Scaling down​

  • Source: Low-traffic period overnight
  • Stimulus: Traffic drops to 10% of normal load
  • Environment: Off-peak hours (02:00 - 06:00)
  • Artifact: All services and infrastructure
  • Response: System scales down resources to reduce costs while maintaining availability
  • Response Measure: No service with more than 2 active instances has less than 10% load for more than 30 minutes

Scenario SC3 - Independent hotspot scaling​

  • Source: Operations team
  • Stimulus: Detects 3 times increased search traffic while checkout is stable
  • Environment: Production, mixed load
  • Artifact: Service-level scaling boundaries
  • Response: System scales only affected services without scaling unrelated ones
  • Response Measure: Bottleneck resolved within 2 minutes; no unnecessary scaling of other components