Google Cloud Well-Architected Framework skill for the Sustainability pillar
面向可持续性支柱的Google Cloud架构完善框架技能
Overview
概述
The Sustainability pillar of the Google Cloud Well-Architected Framework
provides principles and recommendations to help you minimize the environmental
impact of your cloud workloads. It focuses on a shared responsibility
model—Google optimizes the sustainability of the cloud, while customers
optimize sustainability in the cloud. By making informed decisions about
architecture, resource allocation, and region selection, you can significantly
reduce your carbon footprint and improve overall energy efficiency.
Google Cloud架构完善框架的可持续性支柱提供了原则与建议，帮助你最大限度降低云工作负载对环境的影响。它聚焦于共享责任模式——Google优化云本身的可持续性，而客户则优化在云中的可持续性。通过在架构、资源分配和区域选择上做出明智决策，你可以显著减少碳足迹并提升整体能源效率。
Core principles
核心原则
The recommendations in the sustainability pillar of the Well-Architected
Framework are aligned with the following core principles:
Promote a culture of sustainability: Embed sustainability into your
organizational governance, connect technical decisions to environmental
goals, and ensure staff have the skills to implement green practices.
Grounding document:
https://docs.cloud.google.com/architecture/framework/sustainability/culture
Align sustainability practices with industry guidelines: Ensure that
your sustainability initiatives are aligned with industry guidelines for
measurement, reporting, and verification, such as W3C Web Sustainability
Guidelines, Green Software Foundation, and Greenhouse Gas Protocol.
Grounding document:
https://docs.cloud.google.com/architecture/framework/sustainability/industry-guidelines
Ask appropriate questions to understand the sustainability-related requirements
and constraints of the workload and the user's organization. Choose questions
from the following list:
Cloud sustainability:
How do you define the boundaries of sustainability responsibility
between your organization and your cloud provider?
How do you leverage cloud capabilities and AI to drive sustainability
outcomes for your broader business operations?
How does your cloud strategy account for the sustainability impact of
your partner ecosystem and multi-cloud environments?
Use regions that consume low-carbon energy:
How do you incorporate carbon intensity into your Google Cloud region
selection strategy?
Optimize AI and ML workloads:
How do you optimize the energy efficiency of your AI and machine
learning lifecycles?
Optimize resource usage:
How do you ensure your infrastructure footprint dynamically matches
actual workload demand?
How do you select and maintain the hardware types used for your cloud
workloads?
What is your strategy for handling non-urgent or compute-intensive
background tasks?
How do you balance the need for high availability and disaster recovery
with sustainability?
Develop energy-efficient software:
How do you ensure your backend logic minimizes unnecessary CPU, memory,
and network activity?
How do you manage the overall efficiency and maintenance of your
codebase for sustainability?
How do you minimize the data volume and processing load that your
application places on end-user devices?
How does your user experience (UX) design contribute to energy
efficiency for the end user?
Optimize data and storage:
What process do you have for managing the environmental footprint of
your data and storage?
Continuously measure and improve:
How do you analyze your carbon data to prioritize optimization efforts?
How is sustainability measurement embedded into your organization’s
governance and culture?
What is your current process for gaining visibility into your
cloud-related carbon emissions?
What proactive steps do you take to remediate identified carbon
hotspots?
Promote a culture of sustainability:
How do you connect individual technical decisions to the organization's
mission and hold teams accountable for results?
How do you ensure your technical and business staff have the specific
skills required to implement sustainability practices?
提出合适的问题，了解工作负载及用户组织在可持续性方面的需求与约束。可从以下列表中选择问题：
云可持续性：
你如何定义组织与云服务商之间的可持续性责任边界？
你如何利用云能力和AI推动更广泛业务运营的可持续性成果？
你的云策略如何考虑合作伙伴生态系统和多云环境的可持续性影响？
使用低碳能源区域：
你如何将碳强度纳入Google Cloud区域选择策略？
优化AI与ML工作负载：
你如何优化AI和机器学习生命周期的能源效率？
优化资源使用：
你如何确保基础设施规模动态匹配实际工作负载需求？
你如何选择和维护云工作负载使用的硬件类型？
你针对非紧急或计算密集型后台任务的策略是什么？
你如何平衡高可用性、灾难恢复需求与可持续性？
开发节能软件：
你如何确保后端逻辑最大限度减少不必要的CPU、内存和网络活动？
你如何管理代码库的整体效率和维护以实现可持续性？
你如何最大限度减少应用对终端用户设备造成的数据量和处理负载？
你的用户体验（UX）设计如何为终端用户提升能源效率？
优化数据与存储：
你有哪些管理数据和存储环境影响的流程？
持续测量与改进：
你如何分析碳数据以确定优化工作的优先级？
可持续性测量如何融入组织的治理和文化？
你当前获取云相关碳排放可见性的流程是什么？
你采取哪些主动措施整改已识别的碳热点区域？
推广可持续文化：
你如何将个人技术决策与组织使命关联，并让团队对结果负责？
你如何确保技术和业务人员具备实施可持续实践所需的特定技能？
Validation checklist
验证清单
Use the following checklist to evaluate the architecture's alignment with
sustainability recommendations:
Cloud sustainability:
The organization embraces a shared responsibility and shared fate
model for sustainability.
AI is used as a catalyst for profitability and resilience to
streamline operations, or sustainability is integrated into the design
process to create positive feedback loops.
Collaborations with sustainable partners are prioritized and
multi-cloud data portability is leveraged, or internal practices align
with recognized global standards like the Green Software Foundation.
Use regions that consume low-carbon energy:
A data-driven policy prioritizes regions with high Carbon-Free
Energy (CFE%) and "Low CO2" indicators, or the Google Cloud Region
Picker is actively used to balance carbon footprint with cost and
latency.
Optimize AI and ML workloads:
Algorithmic needs are matched to specialized hardware (TPUs) to
maximize computations per watt, or mathematical techniques like model
compression and PEFT are applied to reduce computational complexity.
Optimize resource usage:
Fully managed services that scale to zero when idle are utilized, or
Horizontal Pod Autoscaling (HPA) and Vertical Pod Autoscaling (VPA) are
used in GKE to prevent over-provisioning.
A formal process exists to upgrade to the newest machine types for
improved performance-per-watt, or workloads are actively matched to
specialized machine families.
Batch jobs are proactively scheduled to run during periods or in
regions with the highest proportion of CFE, or Spot VMs are utilized for
non-critical batch jobs.
"Cold DR" or serverless failover is prioritized to ensure secondary
regions remain at zero energy consumption until an event occurs, or
Infrastructure as Code (IaC) is used to rapidly provision a recovery
environment only when needed.
Develop energy-efficient software:
Resource-intensive busy loops or constant polling are replaced with
event-driven logic, or algorithms with optimal time complexity and data
structures are prioritized.
The "Don't Repeat Yourself" (DRY) principle is adhered to with
regular refactoring, or intelligent caching (e.g., Memorystore) is
implemented with smart eviction policies.
The download size of website products is measured and maintained
against a strict budget, or CI/CD pipelines automate the minimization
and compression of HTML, CSS, and JS files.
Static sites or Progressive Web Apps (PWAs) are preferred for faster
loading, or DOM manipulation is minimized to reduce device power
consumption.
Optimize data and storage:
Object Lifecycle Management is used to automatically move cold data
to Archive storage, or discovery techniques (e.g., Dataplex) are used to
identify and eliminate "dark data".
Continuously measure and improve:
Carbon data is analyzed by project, region, and service to identify
gross emitters, or carbon data is joined with Billing data in BigQuery
to correlate cost and environmental impact.
A formal GreenOps function defines accountability for carbon
reduction targets, or verified Carbon Footprint data from BigQuery
supports formal ESG disclosures.
Applications are instrumented to measure the specific carbon
intensity of software features, or automated exports of Carbon Footprint
data to BigQuery are configured for deep analysis.
The unattended project recommender and Active Assist are regularly
used to decommission idle resources, or proactive projects re-architect
hotspots by shifting workloads to low-carbon regions.
Promote a culture of sustainability:
Abstract carbon metrics are transformed into tangible progress
indicators in annual reports, or sustainability is treated as a
first-class technical requirement (NFR) tied to KPIs and performance
reviews.
Training tailored to specific job roles (e.g., developers on code
efficiency, FinOps on carbon unit economics) is provided, or teams are
formally trained to access and interpret carbon footprint data.
使用以下清单评估架构与可持续性建议的契合度：
云可持续性：
组织采用可持续性的共享责任和共享命运模式。
AI被用作提升盈利能力和韧性的催化剂以简化运营，或可持续性被整合到设计流程中以创建正向反馈循环。
优先与可持续合作伙伴协作，并利用多云数据可移植性，或内部实践与绿色软件基金会等公认全球标准对齐。
使用低碳能源区域：
采用数据驱动政策优先选择高无碳能源占比（CFE%）和“低CO₂”指标的区域，或主动使用Google Cloud Region Picker平衡碳足迹与成本、延迟。