⭐1
Identify and describe key sustainability drivers influencing software engineering practicesand decision-making.
- Reduces environmental impact, e.g. through energy efficient code, resource optimization, and mindful infrastructure use.
- Supports economic sustainability, e.g. by ensuring cost efficiency, scalability, and longevity of software systems.
- Promotes social sustainability, e.g. by enabling accessibility, fairness, and positive societal outcomes.
- Enhances technical sustainability, e.g. through maintainable, adaptable, and resilient architectures.
- Respects individual sustainability, e.g. by considering developer well-being, ethical responsibility, and the long-term usability of the system
⭐2
Explain and compare concrete “green” software tactics and design patterns that promote energy efficiency and resource optimization.
CLUE
Cloud-native Sustainability Evaluator
Architecture Styles for Sustainability
Patterns for Sustainability
Implementation Strategies实现
Operational Strategies运维
⭐3
Critically evaluate trade-offs between sustainability goals and other software quality attributes such as reliability and maintainability.
Sustainability Tactic
Gain
Trade-off
| Sustainability Tactic | Gain | Trade-off | | --------------------- | ------------------- | ---------------------- | | Caching | ↓ energy | ↓ consistency | | Serverless | ↓ idle energy | ↑ latency (cold start) | | Microservices | ↓ over-provisioning | ↑ complexity | | Autoscaling | ↓ waste | ↓ spike performance | | Smaller ML models | ↓ energy | ↓ accuracy |
2.1
Architecture Styles for Sustainability
- Monolith vs Microservice,
Monolith:internal network communication->能耗↓; ❌scaling,maintenance,ecolution↑
Microservices:fine-grained scaling;❌network + orchestration overhead - Serverless, [低频请求 + 长 idle period]
scale-to-zero,极低 idle energy;❌cold start，延迟 - Edge/Fog Computing
network traffic能耗↓->network energy & latency;
❌部署复杂度ollout of physical devices、
运维成本Orchestration overhead编排开销
observability可观测性more difficult
Serverless 降低了空闲资源消耗，Edge/Fog Computing 通过减少数据传输来节省网络能
2.3
Implementation Strategies实现
- Energy-aware and Energy Efficient algorithms
-sorting algorithms/- Image compression/- ML-model complexity - Runtime Improvement
versions/ distributions/ libraries or frameworks
new != more efficient; compatibility issues兼容 - Choosing / Changing Language(1step->)
Faster ≠ Greener
code maintainability; ecosystem(框架，tools); developer productivity
2.2
Patterns for Sustainability
- Caching of Read Requests（缓存
节能来源：减少 DB / network I/O
Trade-off：❌ Consistency ↓(–energy efficiency) - Adaptive Service Composition
非必要服务在低负载时关闭
❌User experience ↓(–energy and cost efficiency)
*Sharing Computing Infrastructure
同一platfrom被不同组织的不同system用
design cloud-native systems; introduces additional vendor-dependencies - Circuit BreakerPattern（熔断器
避免 retry storm / cascading failures
本质：resilience tactic → 间接降低无效计算
Implemented in Proxy / Gateway
Re-establishes circuit after some time
2.4
Operational Strategies运维
- Temporal Load Shifting
将 shiftable workload 安排到 renewable-rich periods 执行，以降低 carbon intensity，而非减少功能 - Spatial Load Shifting
通过将 workload 动态迁移到 regions with high renewable energy rates 来降低 carbon footprint，同时需权衡 latency 以及 regulatory and privacy related restrictions - Autoscaling & Right-Sizing
Autoscaling：根据 workload 动态扩展或收缩资源实例数量，以减少 idle resources 并在负载变化时保持系统可用性。
Right-Sizing：匹配 resource capacity 与实际需求，避免 over-provisioning (资源恰当配置) - LightSwitchOps
在downtime periods自动switching off idle resources(VM），以直接降低 idle energy consumption 并减少浪费
idle
| PPT 术语 | 实际意思 |
——————– | —————– |
| idle resources | 分配了但没干活的算力 |
| scale-to-zero | 把 idle 压到 0 |
| over-provisioning | 主动制造 idle |
| right-sizing | 减少 idle |
| LightSwitchOps | 定时消灭 idle |
| cold start trade-off | 用 latency 换少 idle |
trade-off
第 29 页 (Monolith): 提到 Monolith 虽然能减少能源消耗，但其 Scaling, Maintenance and Evolution（扩展、维护和演进）可能耗时更长（即与可伸缩性和可维护性的权衡）。
第 36 页 (Caching): 明确指出 “We trade consistency for energy efficiency.”（我们用数据一致性换取能源效率）。
第 37 页 (Adaptive Service Composition): 明确指出 “Customer experience suffers for the sake of energy and cost efficiency.”（为了能源和成本效率，用户体验可能会受到影响）。
第 51 页 (Right-Sizing): 提到其风险在于如果工作负载或使用模式意外变化，可能影响 availability（可用性）和 stability（稳定性）。
第 61-62 页 (ADR 结果): 总结了迁移到微服务（为提高能效）的负面后果，包括 Increased Complexity（增加了复杂性）、Deployment Overhead（增加了部署开销）、Migration Cost（迁移成本），以及明确的 Energy Trade-off（能源权衡，指服务间通信和容器编排的能源增加）。
ADR
Context – Background and forces leading to
the decision
Decision – The option chosen
Alternatives – Rejected or deferred options
Consequences – Trade-offs, risks, and expected impacts
Status – Proposed / Accepted / Deprecated
