12 - Design Tradeoffs

Well-Architected Framework tradeoffs and decision guidance

---

⚖️ Overview

The Azure Well-Architected Framework acknowledges that design decisions involve tradeoffs between the five pillars. This document captures the key tradeoffs specific to Azure API Management.

---

🏢 High Availability vs. Cost

WAF Tradeoff: Adding redundancy affects costs

Decision Matrix

Requirement	Configuration	Est. Monthly Cost	SLA
Basic Production	Premium 1 unit	~$2,800	99.95%
Zone Redundant	Premium 3 units	~$8,400	99.99%
Multi-Region (A/P)	Premium 3+1 units	~$11,200	99.99%+
Multi-Region (A/A)	Premium 3+3 units	~$16,800	99.99%+

Considerations

• Zone redundancy: Requires minimum 3 units for full zone coverage
• Multi-region: Adds operational costs for failover coordination
• Backend coordination: DR must align with backend failover strategies

---

🏗️ Isolation vs. Operational Complexity

WAF Tradeoff: Isolating workloads adds operational complexity

Approaches Comparison

Approach	Isolation Level	Cost	Complexity
Single Instance	None	$	Low
Workspaces	Logical (RBAC, network)	$$	Medium
Separate Instances	Physical	$$$	High

When to Use Each

Scenario	Recommended Approach
Small team, low risk	Single Instance
Multi-team, shared costs	Workspaces
Strict compliance/data sovereignty	Separate Instances
Maximum blast radius isolation	Separate Instances
Cost optimization priority	Workspaces

---

📈 Scale to Match Demand

WAF Tradeoff: Autoscaling handles malicious traffic too

The Dilemma

Mitigation Strategies

Strategy	Benefit	Cost
WAF	Block malicious before APIM	WAF costs
DDoS Protection	Block volumetric attacks	DDoS tier costs
Rate Limiting	Cap per-client requests	None
Scale Limits	Cap maximum units	Reliability risk

Recommended Configuration

resource autoscale 'Microsoft.Insights/autoscalesettings@2022-10-01' = {
  properties: {
    profiles: [{
      capacity: {
        minimum: '2'   // Baseline
        maximum: '8'   // Cap to control costs
        default: '2'
      }
      rules: [
        // Scale out for legitimate load
        {
          metricTrigger: {
            metricName: 'Capacity'
            threshold: 70
            operator: 'GreaterThan'
          }
          scaleAction: {
            direction: 'Increase'
            value: '1'
            cooldown: 'PT10M'
          }
        }
      ]
    }]
  }
}

---

🔄 Federated vs. Distributed

WAF Tradeoff: Colocation vs autonomous topology

Federated (Workspaces)

Pros	Cons
✅ Cost sharing across teams	❌ Shared outage blast radius
✅ Centralized governance	❌ Misconfiguration impacts all
✅ Economies of scale	❌ Complex multi-tenant RBAC
✅ Single control plane	❌ Capacity planning for all

Distributed (Separate Instances)

Pros	Cons
✅ Full isolation	❌ Duplicative costs
✅ Independent scaling	❌ Redundant operations
✅ Team autonomy	❌ No cost sharing
✅ Blast radius mitigation	❌ Multiple control planes

Decision Guidance

---

💾 Caching Tradeoffs

WAF Tradeoff: External cache can introduce failure points

Caching Option	Performance	Reliability Risk	Cost
No Cache	Baseline	None	$
Built-in Cache	Improved	Minimal	$
External Redis	Best	Additional dependency	$$$

When to Use External Cache

• Built-in cache capacity exceeded
• Cache data > 50 MB (tier dependent)
• Need for advanced Redis features
• Multi-region cache consistency required

Mitigation for External Cache

<!-- Graceful degradation if cache fails -->
<inbound>
    <cache-lookup caching-type="external" 
                  timeout-in-seconds="2"
                  must-revalidate="false" />
    <on-error>
        <!-- Proceed without cache on Redis failure -->
        <set-variable name="cache-failed" value="true" />
    </on-error>
</inbound>

---

📊 Summary: Tradeoff Matrix

Decision	Option A	Option B	Primary Tradeoff
Tier Selection	Premium	Standard v2	Features vs. Cost
Redundancy	Zone/Multi-Region	Single Unit	Reliability vs. Cost
Isolation	Workspaces	Separate Instances	Cost vs. Blast Radius
Scaling	Unlimited	Capped	Reliability vs. Cost
Caching	Built-in	External Redis	Simplicity vs. Performance
Gateway	Cloud-hosted	Self-hosted	Simplicity vs. Latency

---

✅ Tradeoff Checklist

• Documented SLA requirements vs. budget constraints
• Evaluated blast radius requirements
• Assessed team autonomy vs. centralized governance needs
• Defined scaling limits to control costs
• Evaluated caching strategy reliability implications
• Considered hybrid approaches for mixed requirements

---

🔗 Related Documents

Document	Description
01-Architecture	Tier selection
02-Reliability	HA patterns
09-Cost-Optimization	Cost strategies

---

📚 References

• Azure Well-Architected Framework - APIM Tradeoffs
• API Management Workspaces
• Multi-region deployment

---

Back to: README - Main documentation index