Most companies underestimate the time and cost of moving to the cloud. During the transition, organizations incur the costs of maintaining both a cloud infrastructure and service stack as well as their legacy, on-premise stack. This hybrid state can add 20-25 percent to the cost of an on-premise stack or cloud stack alone.
That said, it pays to minimize the time spent in this transitional state. Once the stack is completely in the cloud, with full functionality and resiliency, you can rapidly achieve cost savings by quickly exiting the physical data center and repurposing or decommissioning any hardware and physical infrastructure.
Point B's Perspective
Point B helps organizations reap the benefits of the cloud and avoid the problems that can make the transition costly and time-consuming. Based on our experience, here are a few rules of the road to make your journey smooth and cost-efficient.
Get executive leadership on board.
We see the greatest successes when executives make cloud migration a priority and commit to the resources it will take. Without strong executive support, transitions can drag on, and the hybrid state gets expensive – even demoralizing. When leaders are fully on board from day one, they can focus attention and resources to make a successful move as quickly as possible. This requires the commitment of the entire C-suite, together with everyone who owns the IT infrastructure and the finance teams supporting them.
Understand what you have.
Companies with mature IT organizations are likely to have a firm grasp of which hardware and software are running in their data centers. However, at the current speed of business change, it's easy for tracking processes to quickly lose track of the actual state of things in a data center.
Launch your migration effort with a solid inventory of all hardware and software components in your data center. Which services are running on which hardware, and in which location? Are there any dependencies between services and hardware? If you don’t have an accurate record of what is running in your data center, invest the time and effort to gather this information before you embark on a cloud migration. Regardless of what it costs to do this, it will more than pay for itself in problem avoidance as you migrate.
With a detailed inventory and dependencies in hand, you can begin to plan for decommissioning. Identify the organizations and individuals needed for the migration, and scope the magnitude of the effort.
One typical approach is to migrate all services that make up a particular set of business functionality. For example, if you have a dozen services (or micro-services) that make up an end-to-end payment process, you would coordinate a migration of all these services along a similar timeline and manage them as a single project.
What stays? What goes?
There are situations in which you do not need to migrate certain functionality to the cloud. For example: if you have a service you’ve developed in-house and the cloud platform you’re moving to already offers comparable functionality (such as scalability), you do not need to migrate that functionality and can deprecate it once you migrate everything that uses it to the cloud.
It may make sense to keep some services on-premise. This is usually due to security considerations, potential performance issues or hardware requirements. Most organizations retain some sort of data center, although much reduced. If you have multiple data center rooms, develop a strategy to consolidate on-premise services into as few rooms or racks as possible.
For each service that will move to the cloud, decide whether you will "lift and shift" the service as it is, with only the modifications required to run in the cloud, or whether you will re-factor the service completely to take full advantage of cloud capabilities for scalability and resiliency.
Create a migration plan.
Once you know what and how you want to migrate to the cloud, develop a migration plan and ensure that your company assigns adequate resources to accomplish the work within the allocated timeline. The plan should address all hardware and services, including their interdependencies and business prioritization.
In many cases, it makes sense to establish a dedicated team to perform much of the migration. Consider bringing on additional engineering capacity to address day-to-day business demands so key people with application and system expertise have ample time to focus on the migration.
Most migrations are executed as a gradual ramp-up of transactions over a period of time, rather than a hard cutover. Begin with a very small percentage of total transaction traffic – typically less than one percent – to prove that transaction success rates are at least as high as those of on-premise services. From there, you can gradually increase transactions to 10 percent, and then 20-50 percent, as long as transaction success rates continue to be adequate. Once stakeholders are comfortable at the 50 percent milestone, you should be able to scale to 100 percent relatively quickly.
As noted, embarking on a program of this scope and importance calls for strong executive commitment, beginning in the C-suite. Leadership must be dedicated to making the program a priority, giving it visible support, and providing the dollars and people to see it through. Leadership can help build momentum and keep the lines of communication open as people take on changing roles and responsibilities.
Success is built on a strong program structure, together with the program and project management to accurately plan and track progress toward the end goal. It takes program leadership to manage the data center decommissioning process, and technical expertise to analyze and plan what can be deprecated when. Encourage transparency to identify any blocking issues and address them if they come up.
The Bottom Line
A data center reduction program is a major undertaking for any company. The larger the organizational footprint, the more challenging it will be. To ensure success, get executives on board from the start. Know what hardware and services you have, and what you'll do with them. Build a reliable migration plan. And execute relentlessly when you're ready to go. The sooner you're able to move out of the hybrid state, the sooner you'll realize the full benefits of the cloud.