Advanced Computing in the Age of AI | Tuesday, April 16, 2024

Keys to a Successful Cloud Migration 

via Shutterstock

No matter what study you look at, it’s clear that cloud adoption is growing rapidly – and increasingly workloads are being shifted from the data center to clouds. Key considerations for planning successful cloud migrations start with a full understanding of the different kinds of workloads that will move and the implications for each.

The workload migration planning process includes:

- Discovery, providing an overall view of the IT assets that will be migrated;

- Workload analysis, to determine technical requirements;

- Mapping and risk mitigation to identify security issues, such as those related to data privacy and compliance;

- Planning and execution.

It is important to allow sufficient time for the migration process. There are many critical moving parts in workload migrations and rushing through any step can impact downstream actions. Be realistic when setting timeframes. Typically, migrations take 30-60 days from start to finish but that, of course, varies depending on the complexity of the workload.

Also, don’t be afraid to ask for help. Usually, cloud providers can offer assistance or recommend partners with expertise to assist with migrations.

Step One: Discovery

This first phase encompasses an overall view of the IT assets that will be migrated to the new cloud environment. This involves a thorough evaluation of the workloads and the components in the application stack. Also, check software licenses and, if necessary, consult with the vendor on appropriate permissions.

Step Two: Workload Analysis

Here, the organization works with its cloud provider to review the functional requirements of each asset, and determine which workloads are the most important and most suitable to move to the cloud. Keep in mind that not every workload should be included in the list of migration candidates. Some workloads are more appropriate to remain on-premises. Some may have technical requirements that prevent a move, while others may simply be too expensive or too low-priority to justify their migration. Determine if there have been customizations in the existing workloads that would impact their ability to run in the cloud – and specify any additional work required, or if it is better to leave them on-premises.

Hostway's Eric Brinkman

It’s important to understand that cloud workloads need to be designed to run on the cloud infrastructure while those on-premises were designed for specific hardware and software configurations – and those may not match. It’s possible, too, that part of the workload moves to the cloud and the other part remains on-premises as a hybrid cloud implementation. Lastly, consider the length of time it will take for data transfer, which can be non-trivial in the case of large-size workloads due to storage transfer rates and costly bandwidth charges.

Step Three: Mapping and Risk Mitigation

Next, organizations and their cloud provider can map which type of cloud each workload will be migrated to – public, private or hybrid. This is also an opportune time to identify and mitigate any risks that come up, including security considerations such as data privacy, security or compliance to ensure data is protected and only accessed by authorized users. Consider, too, requirements for availability, response time and backups. This step ensures that project stakeholders can validate their efforts and make streamlined decisions about which workload candidates will make the cut and be migrated.

Step Four: Planning and Execution

After any final obstacles have been addressed with the help of the cloud provider, workloads can begin to be migrated. As noted, it's helpful to migrate items according to their workload classifications and sources. For example, workloads on physical servers can all be migrated together, as can databases on virtual servers. This ensures that each type of asset – including applications, databases, operating systems and underlying infrastructure – is properly organized and reaches the correct cloud environment.

Eric Brinkman, director of product technology, Hostway

EnterpriseAI