Calculating cloud migration costs: What CIOs need to consider

Pre-migration costs: Planning isn't free

CIOs should have a clear idea of how much it will cost to prepare for the cloud migration. At a minimum, it will consume a significant amount of staff time. Migration teams typically center around service or application managers and associated systems staff but will tap resources in storage and network engineering and security. Depending on the application within an organization, migration can also require input from integration specialists, risk management teams and development teams.

Beyond staff time, CIOs should be prepared to budget for the following cloud migration costs:

• hiring or staff development to prepare employees, or professional services to help move a system if their own staff resources are insufficient in numbers or in skillset;
• assessment tools to identify workload interdependencies;
• assessment and provisioning tools to determine how to provision the compute, storage, networking and security components for the workload in the cloud environment;
• new management tools that are better able to deliver the status and troubleshooting information needed for ongoing operations, as it's best to have these on hand for the first migration instead of backfilling them after moving the production environment;
• changes in application and platform costs based on hosting in the cloud, as operating systems, databases and other middleware, and packaged applications may have different licensing terms when running in the cloud than on premises -- and it may take more, or fewer, licenses when running in the cloud; and
• adding cloud access networking from the data center to the destination environment -- e.g., direct connect or a connection through a cloud exchange -- before the migration begins if the migration period will stretch over months, or if there is going to be any significant leave-behind infrastructure in the data center once migrations are complete.

Cloud migration costs

The migration itself will also consume staff time from the same folks who do the planning, whether it's to perform the actual cloud migration or after the fact to verify that the migration is successful. Post-move verification will be the most time-consuming the first few times through. As the team gains experience and the optimization of configurations and automation is complete, the need to pull in network engineers to verify that the most recently migrated workload is running in a properly configured network environment becomes rarer.

Beyond that staff time, CIOs should plan to budget for the following cloud migration costs:

• spending money on systems in two places at once -- for example, supporting an on-premises VDI farm at the same time as a cloud-based VDI plant for as long as there are still on-premises apps requiring VDI access, or paying the platform cost for a big application server while applications migrate out of it;
• maintaining business continuity and disaster recovery (BCDR) spending for on-premises systems until well after production shifts to the cloud; and
• possibly needing professional services or new software to manage actual migrations.

Post-migration: Estimating cloud costs

On average, enterprises spend about 12% more to run a workload in IaaS than they do running it in their own data center. That average encompasses both workloads on which they manage to achieve great savings and those on which costs double or triple when compared to the cost of providing the service themselves.

The more work that is lifted and shifted without modification, the more likely it is that the CIO will have to budget for cost increases.

There are a variety of tools -- available from major IaaS providers as well as from third-party cloud service brokers and cloud cost management vendors -- for estimating the compute, storage and network costs associated with a planned cloud environment. However, to know whether to expect to spend more or less on cloud migration costs, CIOs must have a clear idea of how much they spend now to run a given workload and how much that cost will decrease when the workload moves.

The key thing to understand is many current costs won't vanish immediately. The cost of powering, maintaining, cooling and operating a storage array or server chassis doesn't go away until that array or chassis is empty -- even if it may dip as utilization decreases. If the components in question serve many workloads, that can take months.

The cost of the staff dedicated to managing storage arrays or server chassis also remains until then. If they're responsible for other chassis or arrays, those costs remain ongoing until the workloads are shut down. The cost of operating the data center only goes away when the data center itself is emptied. And cloud ops is not free; skilled staff are still essential, as there may be a new array of provisioning, management and monitoring tools.

CIOs who want to plan for a cloud migration, even if they are not doing so in order to save money, must take care to model cloud migration costs not just for what they will spend in the cloud environment itself, but also for the planning and migration phases and all the costs of operating the cloud environment once it is established. Failing to do so can invite unpleasant surprises for the rest of leadership -- including the CEO and the CFO.