Focus on the most valuable applications and datasets
A key part of cloud strategy is evaluating which applications to move to the cloud.
To do that, we have extended our application rationalization approach described in a previous blog “Six Steps to Rationalizing Your IT Portfolio.”
In our approach, we use surveys and interviews with business and IT owners to place each application on a tolerate, invest, migrate, and eliminate (TIME) quadrant:
- Applications in the tolerate category are adequate for business needs. They have a low business value but a high technical value.
- Applications rated as invest have high business and technical values and are mission-critical applications whose functionalities are vital and need to be modernized.
- The migrate quadrant (high business value and low technical value) contains applications with needed capabilities that can be moved to a new platform.
- Applications in the eliminate quadrant have low technical and business values. They are no longer used or no longer needed and can be retired.
Find the best migration option for each application
Cloud migration only concerns applications that are located in the invest and migrate quadrants (the ones with high business value). The next step is to determine the type of cloud migration. There are different options for migrating applications to the cloud that require increasing modifications to the code and architecture and, therefore, increasing cost:
- Retain. Do not move the application to the cloud and leave it on premises.
- Rehost (Lift and Shift). Deploy the application as-is on a virtual server in the cloud.
- Replatform (Lift-Tinker-and-Shift). As the application is deployed to a virtual server in the cloud, leverage some cloud optimization features by replacing some application modules with open source equivalents or offerings from the cloud service provider (e.g., cloud database, cloud application server).
- Refactor. Re-architect or rebuild applications to leverage cloud-native capabilities. Some of the original code is retained and fit into a different application architecture.
- Replace. Migrate to a completely different product, often software as a service.
Each option requires applications to have some characteristics to be effective. For example, an application that is built around well-defined services is an excellent candidate for the Replatform option, while a monolithic application using old technology is not.
Our interview guides for application rationalization include a series of questions that help determine what migration options apply to a particular application. Based on the answers to these questions, our algorithm suggests applicable migration options and ranks them based on an analysis of alternatives.
Choosing an option that requires limited code changes and no modification in the architecture is less expensive but may not yield all the expected benefits. Often applications migrated using a simple Lift and Shift method failed because bandwidth requirements for access by and interface with other applications are not considered when planning the migration. On the other end of the spectrum, an organization does not have the resources and funding to refactor all its applications. The organization needs to find a midway point that balances cost and benefit. The ranking of options is the starting point to find this balance.
Plan the overall migration, execute the plan, and replan as required
Each application migration effort can be seen a project that competes with other projects for limited funding and resources. These projects need to be prioritized and sequenced based on the directions set in the organization’s cloud strategy and available funding. The key output of this effort is a migration plan that indicates whether, how, and when each application will be migrated to the cloud and how much it will cost at each step of the process. It is then the role of the governance structure to track the implementation of this plan and make the appropriate adjustments as priorities change or problems arise.
LMI assists agencies across the federal government in developing cloud migration strategies to gain the greatest cost savings and efficiencies.
Dr. Perdu has more than 20 years of experience in development and analysis of EAs. He has supported numerous EA projects with several different federal agencies, including GSA, GPO, and the U.S. Army. He holds a PhD in information technology from George Mason University and an MS in technology and policy from MIT.