A Practical Guide to Cloud Migration

Chapter 1. Managing a Successful Transformation

Andrew Gold, Strategic Cloud Engineer

Q: How do you eat an elephant?

A: One bite at a time.

Who

• CxOs

• Technical Leads

• Managers

• Enabling Team Leads

• Engineers

Why

While one can argue that the ultimate goal of digital transformation is to change a “normal” company into a technology company, it is important to recognize that transformation takes place in gradual phases, in accordance with operating pressures and business needs. The immediate goal of transformation should be to produce a happier, more efficient, more effective, and more profitable company. By any reasonable standard, the process should make an organization better. A successful transformation should be good for everyone in the organization, from the CEO on down. Admittedly, a successful cloud migration is unlikely to make the janitor’s task of cleaning office cafeterias any easier, but it is safe to say that nothing good happens to employees of unsuccessful companies.

Let’s consider some thoughtful guidance for the reflective organization: achieving success is not a given, but there are very real rewards for those willing to earnestly strive toward the goal. The leadership team will be able to demonstrate greater efficiencies and lower costs, thus displaying its competence to shareholders. The members of the technical team have a rare opportunity to refresh their skills and make themselves more marketable. The sales teams will benefit from more competitive offerings, leading to greater market share and the corresponding rewards. Even service personnel, like the janitor mentioned above, will enjoy greater job security and a more positive working environment, as successful organizations offer more opportunities and better benefits to employees at all levels.

In a business technical context, a successful digital transformation and cloud migration should lead to lower fixed costs, a more flexible allocation of resources, faster development, worldwide reach, a safer change control process, and faster time-to-market with objectively better products. What’s not to like?

Before an organization begins a migration to cloud, it must have a workable end-to-end strategy for large-scale transformation. For a strategy to be workable, it must be simple, direct, and effective. If the basic strategy cannot be communicated to a roomful of stakeholders in under 10 minutes, something is badly wrong. Sharing the basic strategy with some of the nontechnical teams can also be helpful, as cloud services can have a significant impact on administrative teams such as human relations, vendor management, auditing, and security, to name just a few.

It’s a mistake to underestimate the difficulties involved in the successful execution of a digital transformation. While such a transformation is well worth doing, and perhaps even necessary for the survival of the organization, the effort behind it will be very strenuous, and success will demand everything the organization is able to give. Before embarking on a journey of this magnitude, you must be certain that the organizational will to achieve these goals is present.

Beyond will, an organization must understand its business and technical drivers and then focus on excelling at the tasks that further those drivers. Technology can be a fashion industry, with trends moving in and out of vogue, methodologies rising and falling, and the leading edge of today becoming the technical debt of tomorrow. Therefore, it is essential that an organization hold firmly to its core values and competencies and not be distracted by the most current technology fads.

For example, retail organizations must offer desirable products to the public at a competitive price, while still retaining acceptable profit margins. A realistic digital transformation strategy focuses on the evolution of these drivers and effectively communicates the necessity and importance of these changes to everyone on the project. For a rapid transformation to occur, an organization must either want or need to change, and the impetus must be very powerful or else cultural inertia will halt or severely degrade any progress.

Transformation is not an all-or-nothing process, and there is no requirement that an entire organization must migrate to cloud in order to enjoy significant benefits. While a clear and direct plan is recommended, the first step could be as simple as moving one application into the cloud, performing a postmortem, and then making a more comprehensive plan based on the lessons learned. It may very well be the case that select services can be moved without undue disruption or excessive effort. By limiting the scope of proposed change to well-understood domains, individual thought leaders can have a visible impact on modernizing their working environment and show a technical path forward to leadership. Three or four individuals taking the initiative to demonstrate the benefits of technical modernization can change the direction of an entire engineering organization. If an organization has significant doubts about the viability of a cloud migration, these single-service migrations can be effective and useful first steps toward a more comprehensive effort.

It’s important to be realistic in the initial planning stages. A very high percentage of major IT initiatives either fail outright or are rescoped downward so that some form of “victory” can be declared.¹ It is a rare cloud migration project indeed in which the principals are not under extreme pressure by the final release phases. Wishful thinking and undue optimism have no place in developing a migration strategy, as major structural transformations are very demanding and require the utmost exertion from the leadership and implementation teams. This is true even for the most competent of organizations, let alone for normal ones. For an organization to transition from an existing state to a more desired state, it must understand where it is, where it wants to be, and the processes that convert from one to the other. This understanding should form the basis of an organization’s transformation strategy.

The most common migration failure mode is trying to do too much too soon on a tight schedule. Little is understood about the new platform, so the organization experiences failure and then gives up. If it had just started small with strategic, low-hanging fruit, the organization most likely would have learned from modest early victories and then moved on with its transformation. One specific technique in this regard is to begin the transformation with only a single early adopter or first mover. Learn from the experience, and only then begin broader planning.

How

Let’s look at the specific teams involved in managing a successful transformation, as well as at considerations for how to approach the transformation.

The Teams

By looking at the teams frequently involved in this transformation, we can get a sense of the organizational landscape. This will also give us insight into each team’s motivation, allowing us in turn to fit them into our cloud migration strategy.

The executive team

If the executive team doesn’t align on the need to transform, there’s usually little point in proceeding, as internal strife will ultimately lead to paralysis, slipped schedules, and increases in cost. If the executive team lacks internal alignment, it won’t be able to successfully mediate the inevitable conflicts among the implementation teams. While there is always room in a serious technical endeavor for differing points of view, everyone involved must honestly embrace the ultimate goals without reservation, even if opinions differ as to the most appropriate means of achieving those goals. Alignment, in this sense, does not mean an uncritical or unthinking acceptance of assertions from senior leadership; rather, it represents an honest intellectual comprehension of a thoughtful and methodical policy. This process is not without effort, as it is very important that objections and counterarguments be given due consideration and honestly rebutted. Once leadership has achieved a directional alignment, the next step is to thoroughly and convincingly communicate the new direction to the entire organization; a solid understanding of the strategic hows and whys will help the implementation teams remain closely aligned with the vision of the leadership team.

Finally, cultural transformation and digital upskilling begin with the executive leadership team itself. Middle and junior leadership will emulate the actions of the executive team, and the implementation teams will emulate their immediate superiors, so the executive team should do what it does best: lead by example. If the executive team can transform itself, the battle to transform the remainder of the organization is nearly won.

The platform consulting team

While doing so is entirely optional, many organizations find it helpful to rely on the technical expertise of the platform vendor and their partner consulting services.

In broad strokes, the role of the vendor professional services team is to provide expert guidance in cloud migration and digital transformation. There is a difference here between the sales pitch, which is intended to persuade, and the operational execution that MUST succeed. Beyond the usual sales boilerplate about “trusted advisors” and “partners in success,” the primary role of the platform team is to adequately prepare an organization for the upcoming effort, guide the organization through the usual series of obstacles and impediments, and challenge the organization to execute the steps necessary for success. In more concrete terms, the consulting team will delineate the project infrastructure and communicate the specific process improvements necessary for change to the organization’s leadership team. The consulting team will also develop a comprehensive program to implement those improvements and do everything within its means to make sure the digital transformation is successful.

Partner consulting teams are typically used to augment the core platform team, which is usually quite small. The vendor partner consulting teams serve two primary functions. The first is helping to train and guide the organization’s implementation teams through the initial steps of platform onboarding, and the second is helping to create the initial platform landing zone. In the first phase, consulting teams will perform most of the hands-on “keyboard” work necessary to jump-start the migration process, but they ultimately will transition to more of an advisory and educational role as the implementation teams gain experience and effectiveness.

The client leadership team has every right to expect excellence in both the vendor professional services team and any partner teams brought onto the project. Any concerns or doubts should be raised immediately with the professional services leadership, as it is their responsibility to ensure operational excellence from the entire consulting team. There is no reason to accept second-class professional services talent from any of the major platform vendors.

It is an anti-pattern for the partner teams to act as the primary implementation teams beyond the first stages of the migration process, except for very specific situations in which exceptional domain knowledge is necessary. Why an anti-pattern? Simply put, undue reliance on the partner teams prevents the in-house teams from gaining the knowledge and platform expertise necessary to take ownership of the project as a whole. The danger is that swift progress will be made by the partner teams while they are present, but paralysis and confusion will occur when they leave, because the in-house teams failed to truly take ownership of key project elements.

The internal implementation teams

The implementation teams must learn as much as possible as quickly as possible. This is both easy to say and hard to do, but it is also the simple truth. The process can be accelerated by having individual teams focus on the specific skills relevant to their normal operations, but a certain body of general knowledge is also necessary if the teams are ever to achieve genuine comfort in the new environment. Structured training has an important role, as do the various online self-study tools. It is essential that the teams embrace the new environment with an earnest desire to acquire new knowledge, as there are no substitutes for enthusiasm and motivation.²

There is a significant learning curve here, and the cloud ecosystems can sometimes behave very differently than the data center environments, so having appropriate expectations is important. The best practices of virtual networking, firewalls, DNS forwarding, and DNS peering, as specific examples, differ profoundly from those in a physical data center, so those new to the platform must educate themselves accordingly.

Cloud platforms are essentially delivered as giant boxes of virtual LEGOs, and, as noted on the outside of a LEGOs box, “some assembly is required.” Conceptually, the platforms can each be thought of as a set of interconnected software toolkits of varying convenience, complexity, and maturity. This is of necessity, as each organization has its own business and technical priorities, and no other approach would support the level of customization required by large enterprises. Flexibility and simplicity are opposing requirements. The platforms do their best to provide both, but as stated in the fine print of a car ad, “your mileage may vary.”

As the teams evolve and become familiar with the new platform and its toolset, they should increase the scope of their responsibilities, first by maintaining the work implemented by the partner teams and then by actively superseding them in new development. Mistakes will inevitably be made, particularly when the work of multiple teams is combined or integrated, but these mistakes should be regarded as learning opportunities and treated accordingly. In the end, the accelerated learning process and the associated technical growth are outright wins for most organizations, and eventually these process improvements will lead to more rapid progress later in the migration. Few things are more satisfying than watching a team that stumbled hesitantly through the configuration of basic cloud services early in the migration gain mastery of the new platform and take control of its own technical destiny.

The individual

In this essay, I focus on driving change in large organizations, but not every organization is large, and it is important to highlight the role of individual technology champions. There is no inherent reason why a cloud migration needs to be guided by external consulting teams, and one can make a very good argument for a gradual migration led by internal teams, the footprint of which expands as the implementation team gains experience and knowledge. This bottom-up approach can also be safer when there is no immediate driving need for change, since the more gradual timeline lowers risk and removes the temptation to choose expediency over principle.

All teams are composed of individuals, and it is sometimes forgotten that team motivation and team leadership evolve from individual motivation and leadership. Every migration effort, large and small, begins as an idea in someone’s head. Empowering individuals really means the active encouragement of excellence. We must be willing to accept ideas that are not our own and to look at problems in a new light when presented with new facts, even when those facts are presented by persons of comparatively low status within the organization. Good ideas can and do come from anywhere if our minds are open and we take the time to listen carefully.

The Methods

So far, I’ve looked at how specific teams are involved in managing a successful transformation, but leadership also plays a key role. Let’s examine leadership, course correction, and time management, which form part of the strategy for working with the teams.

Leadership

A true digital transformation begins with a cultural transformation. As an organization’s culture changes, the internal processes and underlying technology will change accordingly since technology is ultimately the external manifestation of an established internal culture. Of course, changes in the internal processes of an organization can also have a profound effect on its culture, so it is important that process and culture reflect and reinforce each other.

An aligned and focused leadership team with strong conflict resolution skills can overcome any number of obstacles. When the executive team is able to communicate its vision with clarity and vigor to midlevel management and the implementation teams, almost anything is possible. Add in a touch of empathy, a healthy dose of patience, and a genuine desire to responsibly decentralize authority, and all of the leadership ingredients are in place for a successful cultural transformation.

Monitoring Progress and Course Correction

At any time, a team member should be able to look at the project status and know whether progress is good, bad, or indifferent. If the current situation is unclear, stop and perform some form of gap analysis. When things go poorly, a blame-free retrospective that focuses on process failures rather than individuals will go a long way toward correcting most problems. There is no shame in failure, as everyone makes mistakes, and there is no such thing as perfect knowledge—but repeated failure for the same underlying reason is to be avoided. For reflective individuals, the goal is always to improve the development and migration process, not to painfully repeat avoidable errors.

In the real world, there are rewards for success and penalties for failure. In sports, one can always look at the scoreboard for objective verification of the current status, and that analogy can be very useful in this context. It is essential that any sustained effort of significance has clear, unambiguous indicators of success and failure.

To summarize: demand operational clarity, have meaningful milestones, aggressively track progress, embrace the process of blamelessly correcting failures, and make sure that everyone knows the current score.

Learning, Training, and Continuing Education

In practice, the most important group to receive enablement is the executive team itself, as improvements in its knowledge serve to empower the entire organization. Most learning plans overlook this idea, but it is worthy of some reflection.

Significant organizational changes require a great deal of upskilling and training. For training and enablement to be successful, there must be both an active learning plan and executive sponsorship. The active learning plan is used to address any concerns raised by a skills gap analysis, and the executive sponsorship is used to give authority and energy to the plan itself, as a plan without sponsorship is unlikely to produce much in the way of results. More specifically, the internal teams will be subject to many demands throughout the migration process, as they must keep the existing infrastructure and applications running while simultaneously attempting to both learn and implement on the new platform. Without executive support, learning will occur only in the margins, and infrequently at that. This reduction in training velocity is very likely to impede implementation velocity later in the project, so deferred technical enablement carries a significant cost.

Having a clear training strategy that focuses on self-service will simplify and accelerate the learning process. Different team members have differing levels of motivation and will learn at different rates, so having several options available can be helpful. Training environments such as Qwiklabs can be very useful for honing specific skill sets. A best practice is to make available a sandbox environment of some type in which team members can experiment with new technologies, and without fear of impacting others. This can serve several purposes. The planning and creation of a sandbox environment is an exceedingly valuable learning experience for the infrastructure team, and it can also be very valuable for the application development teams, who can then create experimental applications on the new platform.

Another useful practice is to make use of structured online courses such as those available from Coursera, and yet another can be to have internal “brown bag” lunch sessions over pizza or sandwiches. A sample lunch topic might be the best way to right size virtual machines, or how to tune data queries to avoid excessive costs. Those who actively embrace the new technologies should be empowered to access more advanced training as they become ready to consume it, as this will frequently inspire emulation by the less self-directed. Since learning is a continuous process, ongoing training should be a permanent feature of any successful digital transformation.

Time Management

Aggressively track progress on a weekly or even a daily basis. A sense of urgency is absolutely necessary for large organizations to make visible progress, but realistic timelines require thoughtful insight and a deep understanding of the real requirements of production-ready services. This deep understanding is usually absent at the beginning of a digital transformation, when so few of the structural processes on the new platform are completely internalized. It is inevitable on a project of this scope and nature that the implementation challenges will be much better understood at the end of the project than at the beginning. This is the primary reason why engaging a platform consulting team is helpful, as one of its main jobs is to guide its clients around the more obvious pitfalls during the initial project phases.

Without digging into the details of the various project management methodologies, it can be safely generalized that the greatest challenge to planning and time management throughout an organizational transformation is the very long list of unknowns. Simply put, the leadership and project management teams do not yet know what they do not know. This suggests that any plan with rigid deliverables and a rigid timeline is unlikely to be realistic or successful.³

General Dwight D. Eisenhower, a man well known for his planning ability, once observed that “plans are useless, but planning is indispensable.” This quote captures the idea that events can render the concrete details of plans obsolete yet also maintains the utility of the planning process and the important information it provides. Any adopted methodology must handle change gracefully, as there will be significant changes throughout the project—very likely many of them. Plan well and thoroughly, but be prepared to adapt your plans as events unfold.

Pitfalls

In this section, we will look at some of the anti-patterns that commonly arise as part of a large organizational transformation, such as migrating to the cloud. The names might not be immediately familiar, although the symptoms may well be.

The Blame Game

When a team member is publicly humiliated for making a mistake, a cascade of negative events takes place, including the following:

• The team member becomes resentful of the public correction. Usually, they try to hide their resentment, but whether or not others perceive it, it still exists.

• The team is incentivized to hide mistakes rather than surfacing them immediately. The mindset becomes, “I won’t let this happen to me.”

• The team is incentivized to shift responsibility elsewhere. The mindset becomes, “It wasn’t my fault.”

• The team becomes defensive and focuses on not being blamed rather than on achieving excellence. The mindset becomes, “I’ll never try to do more than I’m sure I can complete.”

All of these negative behaviors tend to slow team velocity and reduce progress. One mistake that is specific to Agile is blaming teams or individuals for incomplete sprint stories. Blame in this context might be something as simple as mild public ridicule, but consider the long-term consequences. In effect, the team has been incentivized to plan very conservatively, and to commit only to work it is absolutely certain it can complete within the current sprint. As teams typically relax, if only mentally, once the committed sprint work is done, they have essentially been taught never to exceed expectations. Remember, the actual effect of a custom or practice may be very different from its stated intent. In these cases, it can be helpful to focus on the actual results in order to see which techniques are most effective.

How should these types of problems be solved? All change begins with introspection, so a thorough and honest review of actual team processes and procedures is a good place to start. Many teams “understand” the concept of a blame-free retrospective but still place blame in practice. In the end, the team management techniques that are actually implemented, not the ones that are merely understood, are the ones that have an effect, so bring in external help and support if necessary. Consider creating pilot programs to test new techniques and surface the most helpful ones, as this can be a low-risk way to experiment with process improvements.

Unrealistic Project Tracking and Time Management

The leadership team must have a clear but flexible vision of the project end state, have coherent priorities, make decisions quickly, and communicate effectively. This sounds very simple, but of course the real world does everything possible to impede, confuse, and obscure. Concrete and measurable goals, thoughtful milestones, and a sane timeline, combined with a sound conflict resolution strategy, will enable the leadership team to communicate its vision effectively and win the confidence of the implementation teams. Intelligent risk-taking proportional to the opportunity should be accepted and rewarded. One specific goal should be to carefully track the rate at which the in-house teams take ownership of key project elements, as this will indirectly track the process of training and upskilling. The leadership team should be aggressively proactive without being intrusive and must be ready to quickly resolve the inevitable conflicts and roadblocks.

Since all projects have limited time and resources, thoughtful milestones are necessary to demonstrate progress and to promote a sense of urgency. It is a truism that a job will expand to consume the time allotted to it, and timeboxing can be a very valuable technique in this context, but it must be noted that highly structured plans composed of fixed deliverables and fixed dates are problematic because of the numerous uncertainties. Team energy, morale, and cohesion can be destroyed by the pursuit of goals commonly believed to be irrational or unobtainable. Dates that may have appeared plausible at the beginning of a project will frequently be perceived as unduly optimistic later in the process.

Letting the External Teams Solve the Hard Problems

A comprehensive, in-depth knowledge of platform architecture and individual service implementations is essential for the in-house implementation teams; without it, they will ultimately fail to live up to their potential. This concern is not theoretical and should be taken very seriously by the leadership team.

For example, in a recent project, the offshore partner team was genuinely talented and exceptionally efficient. The client would ask for a feature, and the offshore team would have it complete within a few days, or within a week at most. The client became entirely too comfortable with this situation, which continued for about six months. When the consulting contract reached completion, it was a very sobering experience for the client technical team to realize that its internal staff had very little idea how to operate, maintain, and extend its shiny new system. Ultimately, the client was forced to sign another statement of work with the partner, the intent of which was largely to help the internal team learn to manage its new platform.

Use the external consulting teams to enable the in-house teams, not to do work for them.

We’re Successful, So Why Change Anything?

Every large organization has elements that are strongly resistant to change, and many times the most successful elements are the most resistant. Their thinking is very straightforward: if things are working now and have worked well for the past 10 or 20 years, why make changes? Certainly change for its own sake is counterproductive, but “change” in the current context refers to embracing a more advanced paradigm that bestows measurable, demonstrable advantages. It is imperative that the leadership team communicates the significance of these advantages so the midlevel and implementation teams understand the purpose, goals, and benefits of the new processes. If the midtier teams are doubtful of the organizational commitment, their uncertainty will lead to inertia, indecision, and the indifferent execution of higher-level policies. The leadership team must actively convince the implementation teams of the clarity of its goals and its commitment to a successful transformation. A clearly defined strategy composed of coherent individual elements describing concrete, measurable results is invaluable here.

The Bike Shed Effect

It is easy to become distracted during a migration effort by the myriad of surrounding details. For an organization to effect real change, it must provide useful products and services sooner rather than later. More prescriptively, it means the initial migration effort should restrict scope to a very lean minimum viable product (MVP) and release that MVP, whether internally or externally, as quickly as possible.

Parkinson’s law of triviality argues that people within an organization commonly give disproportionate weight to incidental issues because they understand them more easily than the much more complex challenges of their primary tasks. It is a mistake to dilute the limited available resources through the pursuit of the nonessential or trivial. Decide what is important to the migration effort and do nothing else.

¹ See “ Pitfalls ” for more details.

² See Chapter 2 for more on this topic.

³ See Chapter 5 for more details on this important topic.