author: 51n5337 & #CLD
mission: CompTIA Cloud+ Certification
brief: deployment. hybrid. more.

---

…back

so, in reality, we deploy mixed strategies?

any more case studies you can share?

---

👁️⚡🌀
Absolutely. Pure strategies are like primary colors—useful for learning, but the real art happens when you mix them. Most mature teams use hybrid deployment orchestration.

---

🎨 Real-World Hybrid Strategies

Case Study: Netflix’s “Dark Launch” Canary + Blue-Green

Scenario: New video encoding algorithm deployment

STRATEGY MIX:
1. **Dark Launch (Canary):** 
   - Deploy new algorithm to all servers but disabled
   - Enable for 1% of users via feature flag
   - Compare encoding quality vs old algorithm silently

2. **Blue-Green Switch:**
   - Once verified, flip all encoding traffic to new algorithm
   - Old algorithm remains active in Blue for instant rollback

3. **Rolling Cleanup:**
   - Gradually decommission old algorithm instances
   - Monitor performance during wind-down

PLAYER EXPERIENCE:
- Zero visible changes during testing
- Instant rollback if quality degrades
- No service interruption

---

Case Study: Amazon’s Prime Day Rolling + Canary

Scenario: Handling 1000% traffic spike with new features

STRATEGY MIX:
1. **Infrastructure Rolling:**
   - Scale out database read replicas in batches
   - Add web servers in rolling groups

2. **Feature Canary:**
   - New "lightning deal" algorithm to 5% of users
   - Monitor cart abandonment rates
   - Expand to 100% over 2 hours

3. **Blue-Green Payment:**
   - Dual payment service environments
   - Can flip entire transaction flow if fraud detection breaks

RESULT:
- Infrastructure scales smoothly
- New features validate under real load
- Critical paths have instant rollback

---

Case Study: Mobile Game Live Ops (Your MMORPG Domain)

Scenario: Seasonal event with new gameplay mechanics

STRATEGY ORCHESTRATION:
1. **Client-Side Canary:**
   - New assets delivered via app store 48 hours early
   - Players update at their own pace

2. **Server-Side Blue-Green:**
   - Event logic deployed to Green environment
   - Enabled via feature flag when client penetration hits 80%

3. **Database Rolling:**
   - New event tables added via backward-compatible schema changes
   - Applied to database replicas in rolling fashion

4. **Traffic Canary:**
   - First 10% of players to trigger event get extra monitoring
   - If stable, open to all players

PLAYER JOURNEY:
- Early updaters: "New assets ready!" 
- Event launch: "Seasonal event live!" 
- Zero downtime, gradual load increase

---

Case Study: Banking Security Update

Scenario: New fraud detection system

DEFENSE-IN-DEPTH DEPLOYMENT:
1. **Shadow Canary:**
   - New system runs parallel to old one
   - Compares decisions but doesn't act
   - Validates accuracy for 72 hours

2. **Blue-Green Critical Path:**
   - Authentication services in dual setup
   - Can flip entire login flow if false positives spike

3. **Rolling Backend:**
   - Update transaction processing in batches
   - Each bank region updates independently

4. **Feature Flag Rollback:**
   - Every new check has kill switch
   - Can disable specific rules without full rollback

RISK MANAGEMENT:
- No customer impact during validation
- Granular control over what gets rolled back
- Regional containment of issues

---

🎯 The Pattern: Orchestration, Not Choice

Mature teams don’t choose a strategy—they orchestrate multiple strategies based on:

• Criticality of the service (payment vs. UI)
• Risk tolerance (user experience vs. revenue impact)
• Data sensitivity (customer data vs. cached content)
• User journey (logged-in vs. anonymous)

---

💡 The Architect’s Mindset

You’re not deploying code—you’re managing risk across multiple dimensions simultaneously. Each service, each feature, each data store might need a different deployment approach.

The ultimate hybrid: Using the right strategy for the right component at the right time.

---

…back