This stage focuses on iterating the portfolio-level view, closing data gaps, and obtaining more data to produce high-confidence migration wave plans for the entire portfolio. 此阶段侧重于迭代资产组合层级的视图，填补数据空白，并获取更多数据，从而为整个资产组合制定高可信度的迁移波次计划。
The stakeholders of this stage are typically a mix of the previous two stages. They include CxOs and senior leaders, migration and platform teams, and IT and Enterprise architects. The key is to augment the portfolio-level data fidelity through iteration and refinement.此阶段的利益相关者通常由前两个阶段的参与者组成，包括 CxO 和高级领导者、迁移与平台团队，以及 IT 和企业架构师。关键在于通过迭代和细化，提高资产组合层级数据的保真度。
Tip提示
For details and guidance, see the relevant section in the Application portfolio assessment guide for AWS Cloud migrations.有关详细信息和指导，请参阅《AWS 云迁移应用资产组合评估指南》中的相关章节。
High-level objectives and actions高层目标与行动
Establish a baseline for the application portfolio and associated infrastructure – Iterate portfolio-level data, building up from the discovery acceleration and initial planning stage, to close gaps and produce a high-level view of the entire application portfolio. At this stage, it is key to refine application-to-infrastructure mapping, usage data, and application metadata attributes. These attributes include ownership, criticality, and the primary function of each application.建立应用资产组合及相关基础设施的基准——在发现加速和初步规划阶段的基础上，迭代资产组合层级的数据，以填补空白并生成整个应用资产组合的高层视图。在此阶段，细化应用与基础设施的映射、使用数据以及应用元数据属性至关重要。这些属性包括所有权、关键程度以及每个应用的主要功能。
Obtain and analyze dependency data (typically by using specialized discovery tooling) – To validate applications and aid the creation of migration wave plans, high-confidence application-dependency data is key at this stage. Dependency data includes communication data, such as volume and frequency of communication between systems, and nontechnical dependencies, such as operational considerations. These dependencies dictate which applications must move at the same time and which applications can operate from different locations.获取并分析依赖关系数据（通常使用专门的发现工具）——为了验证应用并辅助创建迁移波次计划，在此阶段，高可信度的应用依赖关系数据至关重要。依赖关系数据包括通信数据（例如系统间通信的容量和频率）以及非技术性依赖关系（例如运营方面的考量）。这些依赖关系决定了哪些应用必须同时迁移，以及哪些应用可以在不同位置运行。
Identify and validate compliance and regulatory requirements – Identify and validate frameworks, rules, evidence, and documentation requirements. 识别并验证合规性与监管要求——识别并验证框架、规则、证据和文档要求。
Convert assumptions into facts – How much has been assumed in previous stages? It is key in this stage to reduce the number of assumptions to a minimum.将假设转化为事实——在之前的阶段中做了多少假设？在此阶段，将假设数量降至最低至关重要。
Establish a baseline for the migration portfolio-rationalization models – Iterate models from previous stages (for example, application prioritization criteria and the 6 Rs decision tree). Validate the models by applying them to the entire application portfolio.建立迁移资产组合合理化模型的基准——迭代之前阶段的模型（例如，应用优先级排序标准和 6R 决策树）。通过将这些模型应用于整个应用资产组合来对其进行验证。
Document measurable business outcomes – To align migration waves to business objectives, identify business outcomes and associated key primary indicators (KPI) for each migration wave.记录可衡量的业务成果——为了使迁移波次与业务目标保持一致，需确定每个迁移波次的业务成果及相关的关键绩效指标 (KPI)。
Evolve the directional business case – Replace benchmarks with actual utilization and cost data, and refine migration costs in line with the updated wave plan. Broaden the coverage to include the full expected value from each expected business outcome, such as cost reduction, IT productivity improvement, increased resilience, and greater agility.演进方向性业务案例——用实际利用率和成本数据替换基准，并根据更新后的波次计划细化迁移成本。扩大覆盖范围，纳入每项预期业务成果带来的全部预期价值，例如成本降低、IT 生产力提升、弹性增强和敏捷性提高。
Document and communicate key dates – Document and communicate compelling events (such as data center exit dates, contracts and licensing agreement renewals), application release cycles, migration dates to avoid, technology refresh cycles, and people availability.记录并传达关键日期——记录并传达重大事件（例如数据中心退出日期、合同和许可协议续签）、应用发布周期、应避免的迁移日期、技术更新周期以及人员可用性。
Analyze cost and risk for wave planning – How much parallel change can be supported or tolerated? Analyze people requirements, risks identification and mitigation, criticality, and impact.分析波次规划的成本与风险——可以支持或容忍多少并行变更？分析人员需求、风险识别与缓解、关键程度以及影响。
Identify skills requirements – What is the projected level of readiness to support the different workload types in the cloud? Does the migration wave plan align to the projected readiness? Can the support teams meet the wave plan requirements?识别技能需求——支持云中不同工作负载类型的准备程度如何？迁移波次计划是否与预期的准备程度一致？支持团队能否满足波次计划的要求？
Document internal processes – Document information about current processes impacting cloud migration, such as change management, service management, architectural review boards, risk assessments, and approval workflows.记录内部流程——记录有关影响云迁移的现有流程的信息，例如变更管理、服务管理、架构审查委员会、风险评估和审批工作流。
Create a migration wave plan – To create a wave plan, combine all the elements previously discussed in this list. Apply prioritization criteria to the portfolio, and analyze dependencies to create waves of applications. Incorporate platform and migration readiness into the migration wave plan. How long does it take to implement AWS infrastructure and services? What is required for security and operational readiness? What is the impact to the wave duration? How long does it take to implement migration tooling? How long does it take to replicate data, considering network and systems usage? What is the cutover window? How long does it take to roll back?创建迁移波次计划——要创建波次计划，需整合此列表中之前讨论的所有要素。将优先级排序标准应用于资产组合，并分析依赖关系以创建应用迁移波次。将平台和迁移准备情况纳入迁移波次计划。实施 AWS 基础设施和服务需要多长时间？安全和运营准备需要什么？对波次持续时间有何影响？实施迁移工具需要多长时间？考虑到网络和系统使用情况，复制数据需要多长时间？切换窗口是多久？回滚需要多长时间？
Update workstreams – Establish a process to feed portfolio and detailed application assessment data into Migration and Landing Zone workstreams. Ensure these workstreams clearly outline their data requirements.更新工作流——建立一个流程，将资产组合和详细的应用评估数据输入到迁移和着陆区 (Landing Zone) 工作流中。确保这些工作流明确概述其数据需求。
Outcomes成果
High-fidelity application and infrastructure inventory高保真应用和基础设施清单
High-level migration strategy for each application每个应用的高层迁移策略
Detailed business case详细的业务案例
High-confidence migration wave plan高可信度迁移波次计划
Best practices最佳实践
Ensure that applications are evenly distributed in the migration wave plan. Consider criticality and complexity to avoid pockets of complexity that could create blockers or delay migrations.确保应用在迁移波次计划中分布均匀。考虑关键程度和复杂性，以避免可能造成阻塞或延迟迁移的复杂性集中区域。
Prioritize noncritical, simple applications in the first two waves.在前两个波次中优先迁移非关键的简单应用。
Focus on combining prioritization, dependencies, and business drivers to iterate the wave plan.专注于结合优先级排序、依赖关系和业务驱动因素来迭代波次计划。
Consider cloud infrastructure, security, and operational readiness (including skills) when creating the wave plan.在创建波次计划时，考虑云基础设施、安全性和运营准备情况（包括技能）。
Construct wave plans so that the length of a migration wave, typically ranging 4-8 weeks, outlines the application journey. Each wave should cover the following:构建波次计划，使迁移波次的时长（通常为 4-8 周）能够概述应用迁移历程。每个波次应涵盖以下内容：
Detailed assessment详细评估
Migration readiness迁移准备情况
Infrastructure build and test基础设施构建与测试
Data transfer数据传输
Cutover of the applications in the wave波次内应用的切换
Wave closure (for example, lessons learned, post-migration issues resolution)波次收尾（例如，经验教训总结、迁移后问题解决）
Define and use a default wave structure to apply a migration factory model that includes detailed assessment, design, implementation, test, cutover, and validation.定义并使用默认的波次结构，以应用包含详细评估、设计、实施、测试、切换和验证的迁移工厂模型。
