Cloud Migration: Strategy and Best Practices Part 3

As organizations increasingly adopt cloud technologies, the process of migrating large-scale workloads to the cloud has become a critical undertaking. However, running a large migration presents unique challenges that requires meticulous planning and execution from the perspectives of people, process and technology. 

​​In the previous two articles on Cloud Migration Strategy and Best Practices, we discussed the importance of having a well-defined strategy, a clear scope, and realistic timeline for successful cloud migration projects. We also highlighted the critical role that people and technology play in the success of cloud migration projects. ​In this article, we will shift our focus to the process perspective of cloud migration in terms of preparing for a large migration and running a large migration.

Process Perspective

When it comes to large-scale migrations, having a well-defined process in place is critical. While processes bring consistency, they must also be adaptable to each project's unique requirements. Running the process repeatedly will help you identify gaps and areas for improvement, leading to significant benefits as you iterate, learn, and adopt new ideas.
​
Managing processes in migrations can be challenging as they often span multiple technologies and boundaries that may not have been linked before. This guide provides processes and guidance on specific requirements for large migrations, helping you achieve success while maintaining quality and team confidence.

Preparing for a Large Migration

To ensure a successful migration journey, it is crucial to establish core principles that provide a clear direction and obtain buy-in from stakeholders.

In this section, we will cover the following topics:
​

• Define business drivers and communicate strategy, cope, and timeline.
• Define a clear escalation path to help remove the blockers.
• Minimize unnecessary change.
• Document an end-to-end process early.
• Document standard migration patterns and artifacts.
• Establish a single source of truth for migration metadata and status.

​
Define Business Drivers and Communicate Strategy, Scope and Timeline

Defining business drivers and having a clear communication plan for the straetgy, scope and timeline are vital for large migrations to AWS. Different migration paths can be considered, such as rehosting workloads, containerizing applications, or redesigning them into serverless architecture. To determine the appropriate migration path, it is important to align with business drivers.

Involving various stakeholders, including application owners, network teams, database administrators, and executive sponsors, is crucial. Documenting business drivers and setting key performance indicators (KPIs) aligned with target outcomes helps ensure stakeholder alignment and effective decision-making.
 
Define a Clear Escalation Path to Help Remove Blockers

Large cloud migration programs involve multiple stakeholders with their own priorities, which can create challenges. To address this, a clear escalation path must be established to outline the necessary actions for removing any blockers that may arise. This streamlines decision-making processes and ensures alignment among teams. An example of resolving conflicting migration paths is setting a clear mandate following an escalation to the Chief Information Officer (CIO) and implementing a mechanism for requesting required decisions.
 
Minimize Unnecessary Change

While change is beneficial, excessive changes can introduce additional risks. When a business case for a large migration is approved, it is recommended to set a two-week rule to prevent application teams from spending excessive time rewriting their applications. This rule helps maintain consistency and enables a sustainable migration process over a multi-year period. By minimizing changes that do not align with the desired business outcomes, mechanisms can be developed to manage such changes in future projects.
 
Document an End-to-End Process Early

Comprehensive documentation of the entire migration process is essential for effective planning. This documentation should assign ownership of tasks and processes to specific stakeholders, ensuring clarity of roles and responsibilities. It also helps identify potential issues and facilitates ongoing improvements.

​Existing processes, dependencies, and integration points should be documented, and a RACI matrix can be created to assign responsibilities and accountabilities. Additionally, establishing a countdown plan, working backward from the workload migration cutover date and time, provides a structured approach.
 
Document Standard Migration Patterns and Artifacts
​
Documenting standard migration patterns and artifacts is critical for success. These resources serve as reusable references, documentation, and runbooks for future migration projects, enabling avoidance of past pitfalls and issues. Standard processes and artifacts significantly accelerate the migration process and improve consistency.

It is recommended to establish central ownership of these documents and artifacts, with a process for submitting recommended changes. Regularly sharing updates and changes with all teams promotes effective communication and ensures consistency throughout the migration project.
 
Establish a Single Source of Truth for Migration Metadata and Status
​
Creating a single source of truth for migration metadata and status is essential for effective planning. This allows all teams to align and make data-driven decisions. Initially, multiple data sources may exist, such as configuration management databases (CMDBs) or inventory lists. Data capture mechanisms, like using discovery tooling or surveying IT leaders, may be necessary. Aggregating all data sources into a single dataset simplifies tracking the migration progress, including the status of migrated servers.​

Running a Large Migration

Once the business outcomes have been established and the migration strategy has been communicated to the stakeholders, it is time to plan how to divide the scope of the large migration into manageable migration events or waves. The following sections provide essential guidance for creating a wave plan.

Plan Migration Waves Ahead of Time to Ensure a Steady Flow

Thorough planning is crucial for the success of the migration program. Planning migration waves in advance allows the project to progress smoothly and enables the team to be proactive in addressing migration requirements. It facilitates scalability, enhances decision-making, and improves forecasting as project demands become more complex. Additionally, planning ahead enhances the team's adaptability to changes. For instance, a financial services customer working on a data center exit program initially faced delays due to sequential wave planning. When stakeholders were informed about their applications' migration to AWS, they still had several tasks to complete before starting the migration, causing significant program delays. To address this, the customer implemented a holistic approach where migration waves were planned months in advance. This provided ample time for application teams to complete pre-migration activities and eliminated unnecessary delays.

Keep Wave Implementation and Wave Planning Separate

Separating the teams responsible for wave planning and wave implementation allows both processes to work concurrently. With effective communication and coordination, this approach avoids slowdowns in the migration caused by insufficiently prepared servers or applications. It is crucial to involve the migration implementation team during wave planning to ensure complete and accurate data collection. Additionally, creating a buffer between wave preparation and implementation is essential. Collaboration between the wave planning team and the migration team is necessary to gather the right data and minimize the need for rework.

Start Small for Great Outcomes

Starting with a small-scale initial wave and gradually increasing migration velocity in subsequent waves leads to favorable outcomes. The first wave should involve a single, small application with fewer than 10 servers. As the migration progresses, additional applications and servers can be included in subsequent waves, gradually building up to the target migration velocity.

Prioritizing less complex or risky applications and incrementally ramping up the migration velocity allows the team to adjust to working together and learn from the process. With each wave, the team can identify and implement process improvements, significantly enhancing the velocity of later waves. For example, a customer migrating over 1,300 servers in a year began with a pilot migration and a few smaller waves.

This approach allowed the team to identify opportunities for improvement, optimize network segments, collaborate with the firewall team to prevent delays, and develop automation scripts for discovery and cutover processes. Starting small enabled the team to focus on process enhancements and increased overall confidence.

Minimize the Number of Cutover Windows

Maintaining discipline in managing scale is crucial for successful mass migrations. Limiting the number of weekly cutover windows ensures that the time spent on cutover activities is maximized. By reducing flexibility in this area, unnecessary delays and operational burdens associated with scheduling are minimized. For instance, instead of having multiple small cutovers, consolidating servers into fewer, larger cutovers optimizes operational efficiency and reduces potential delays.

A large technology company experienced delays early in their migration project due to application teams having the flexibility to dictate migration schedules until the last minute. This resulted in constant negotiation and stress for the migration team. To address this, the company improved their planning discipline and reduced the number of cutover windows, avoiding delays in meeting data center contract expiration dates.

Fail Fast, Apply Experience, and Iterate

It is common for initial migration waves to encounter challenges and setbacks. Failing early in the process allows the team to learn, identify bottlenecks, and apply lessons learned to subsequent waves. During the initial stages of a migration, the team needs time to adjust, integrate various tools and people, and continuously improve the end-to-end process. Understanding and communicating that initial issues are expected is crucial, as some teams may be reluctant to embrace new approaches and failure.

Ensuring everyone understands that these challenges are part of the journey encourages the team to learn, adapt, and ultimately achieve a successful migration. For instance, a company planning to migrate over 10,000 servers in 24-36 months began with learning waves to understand the processes and permissions involved. Through iterative improvements, such as integrating with CMDB and CyberArk, the team increased their migration velocity to over 120 servers per week within six months.

Don't Forget the Retrospective

Conducting retrospectives is an essential part of an agile process. These sessions allow the team to reflect, discuss, learn, and make necessary adjustments before moving forward. Retrospectives provide a structured approach to capturing lessons learned, which can then be used to drive improvements. For large migrations to succeed, constant evolution and improvement of processes, tools, and teams are vital. Retrospectives play a significant role in this continuous improvement cycle.

Instead of waiting until the end of the program, lessons learned from previous waves should be applied to the planning of subsequent waves. Regular retrospectives provide opportunities to identify areas for streamlining, process improvements, and automation. By implementing a countdown schedule and automating manual tasks, one customer significantly minimized delays and optimized their cutover process.

Another large tech company held regular retrospectives that led to improvements in processes, scripts, and automation, resulting in a 40% reduction in average migration time over the course of the program.

Conclusion

Large migrations present different challenges when compared to smaller migrations. This is mostly due to the complexities introduced by the scale. Running a large migration requires meticulous planning and execution as well as effective coordination, and a focus on continuous improvement. By following the guidance provided in this three-part series, organizations can navigate the complexities of large-scale migrations and achieve successful outcomes.​