Companies don't move to the cloud because they want to spend more money. So how should a smart organization approach cloud migration?
Any technical challenge can be overcome if you throw enough money at it, but that's not a sensible cloud migration strategy. Let's look closely at what to move -- and how to move those things. Plus, it's important to consider how to handle a staff that needs to learn about cloud computing. When you take these things into account, your organization will be in a better position to take advantage of the cloud without spending a fortune on the way there.
What to move to the cloud
Business leaders can be good at pointing an organization in a particular direction. "We need to move to the cloud," for example, is sensible and easy enough to say. For those who don't work in IT, though, it's not easy to understand the nature of a multiserver application. To shift that type of app to the cloud without a proper design becomes a 1:1 or 1:more server-migration scenario. That sort of arrangement is unlikely to save money.
IT decision makers will likely need to help non-IT teams learn about cloud computing so that expectations can be managed and details can be worked through. Operating in the cloud, after all, can be expensive -- especially if done haphazardly.
Whether you move to IaaS, SaaS or somewhere in between, you'll need to take a close look at what you're using in the application stack. Oftentimes, on-site applications contain a lot of features that were never deployed or went unused in your environment. What you don't want to do is pay to both move and support those unused features in the cloud. Reporting features, analytical servers and other add-ons to your primary applications can cost thousands in the cloud because you still pay for them even if you don't use them. This can add up.
The other concern with a cloud migration is moving stale data. Are you moving all of your historical data to the cloud? Is there a reason to do this? Is there data you rarely or never access? Even with different storage tiers available, it costs money both to move and store that data in the cloud.
These additional fees will needlessly drive up overall costs. The problem is that the expenses will look tiny to start. A four-core 16 GB RAM server costs $0.22 an hour on demand on AWS; doubling the server to eight cores and 32 GB of RAM would cost $0.44. Run the math out over a full year, and you would spend $1,972 in the scenario and $3,854 in the second; that starts to make a serious difference.
How to move to the cloud
Transferring 100 TB of data on a T3 internet connection would take 269 days, while a 1 Gbps connection -- both up and down speeds -- would accomplish that in 12 days, according to AWS estimates. Consider, though, that 100 TB is simply not that much data anymore. Uncontrolled email boxes can easily grow to become multi-terabyte monsters, and that's just the start.
Sidelining your internet connection for weeks or months is not realistic, and the sync process for changed data once you get it to the cloud is likely to add additional weeks.
This means you need to think about trucks instead of internet transfers. Services such as Snowball and Snowmobile from AWS involve physically shipping the data on secure drives. Microsoft Azure and Google Cloud Platform offer similar methods for offline data shipment. While this could seem like an oddly low-tech way to move to the cloud, the cost is a fraction of what you pay for additional bandwidth. Plus, you'll see a much quicker turnaround time. This won't eliminate the need for additional syncing, but it will reduce the amount of time and expense required.
Staffing up and down for the cloud
Things can get sensitive when the cloud migration discussion turns to personnel. You'll need to prepare the IT team for how their duties will change.
For example, if you change from on-premises email to cloud-based email, it is tempting to assume you'll no longer need to pay an Exchange administrator. A move to the cloud, however, doesn't always cut your requirements for technical staff in that area. The cloud model could change some of the pieces that the staff supports, such as the infrastructure and back-end services. But the interfaces, including the ones that support the authentication from the cloud, need to be managed. Your cloud provider isn't going to do this for you.
Overall, a cloud migration removes some aspects of a traditional IT admin's role, such as hardware support. Migration, however, won't remove that position entirely. Cloud workloads need to be managed. Previously, when the IT team needed to adjust or add additional resources to on-site applications, this wasn't really a problem; you could simply schedule downtime and do what you needed to do. In the cloud, all of those adjustments will cost you, as nothing is free in the cloud.
Also, keep in mind that there will be intricacies about cloud computing that IT staffers will need to learn. The cloud's interconnections can be complex, so you can't simply wing it and hope for the best. Training classes can cost $3,000 to $4,000. These are expenses you'll want to consider when you your move to the cloud and construct your migration budget.
After the move to the cloud
After cloud migration, you will need to address performance, stability and usage. While you can for these to be minor matters, in reality, you will need both policies and procedures for your staff on how to address these issues. This might involve a combination of monitoring tools that require updates or tools to be integrated so that your environments can be properly monitored.
While these might seem to be normal steps, they need to be budgeted. Otherwise, it's an added expense. And, once you're up and running in the cloud, you will need to monitor and contain your spending so that migration doesn't look like a blunder.
Moving to the cloud is a goal that's easy to set, but difficult to pull off. An organization will need to think through exactly what cloud migration might mean. The cloud, after all, is an environment that you have to pay for. Deploying complex, modern applications in that environment will most likely cost more than expected.