Not all cloud migrations are successful, and before you move your workload back to your own data center, there are things you need to consider.

Not all cloud migrations are successful, and before you move your workload back to your own data center, there are things you need to consider.

Moving From Cloud Back to Data Center – Not as Easy as You May Think

Add Your Comments

What if you went down the cloud path and realized it was a mistake? What if you deployed a massive data point or critical application into a cloud data center and it’s not working very well. Now what? How do you pull back and place your environment back into your own data center? Regardless of your organization’s size, migrating critical applications and workloads can be a bit scary. But it doesn’t have to be. Many IT shops have had to move large workloads between public and private cloud instances. In some cases, it simply has to be done, whether it’s for performance reasons or just use-case issues.

Let’s take a look at some of these use cases and what organizations have done to migrate their cloud workloads:

  • Finding the issue. There needs to be a reason. It’s sometimes as easy as that. Many organizations follow trends or adopt technologies without understanding the impact on their existing business model. So what happens if you’re in a situation where a piece of your cloud-based environment just isn’t working the way it needs to? Before ripping everything out, find the issue at hand. Is it a specific application or possibly a resource dependency? Did a piece of your platform greatly outgrow your own expectations? Organizations looking to migrate back to the data center should do so cautiously. In many situations, ripping everything out and making it private again is actually yet another mistake. Taking time to identify a cloud challenge will not only help you understand which piece of your infrastructure should be moved, it’ll also help you proactively plan out your platform.
  • Planning the migration (always try to do this in a parallel deployment). In my numerous experiences and architectural projects, one of the best ways to migrate a system, application or platform is to do it in parallel. Even in large organizations emotions can get the best of people, so creating the least disruptive migration plan is important. You can plan for transitional phases, you can test-dev the system, and you can effectively train your users and your staff as needed. A good architecture can take a single application – or an entire platform – and migrate it in parallel to your existing systems. Planning metrics around capacity, bandwidth and other resources must be taken into consideration. 
  • Replicating data points, resources and users. Just because it works one way in the cloud does not mean it’ll work the same way in a private data center. Take the time to understand the impact of the migration. There are tools out there, like LoginVSI, which can create “shadow” users to test against a system. Can you handle the additional workloads? Do you have the resources needed to scale? Will the application you just migrated work optimally in a new setting? Proactively answering these kinds of questions will ensure a smooth migration process. Storage, networking, compute and the end-user experience must all be considerations. Fortunately, data center and infrastructure technologies have really come a long way. Our ability to replicate data and networks and span the user are much further advanced than ever before. Consider using these tools.
  • Creating the new use case. Now that you’ve elected to migrate a piece of your environment, it’s important to create your new use case. Why? Because you don’t want to have the same challenges more than once. Deploying a resources on a system requires planning and a good use-case scenario. The application which was functioning in a public cloud environment will operate and interact differently when you bring it in-house. Even before you migrate something from a public cloud environment, you need to create a good use case, because there may be a solid chance that this same resource also won’t work well in a private data center.
  • Ensuring workload longevity. Workload, application or even infrastructure migration is never easy. In today’s ever-evolving technology world it’s critical to ensure longevity of your data and workloads. During the migration process (and after) take the time to plan out your resources and requirements. Scalability and supporting the end-user from a performance perspective will go a long way in allowing your platform to run well. Look for optimizations to help. There have been ltos of innovations in the software-defined world. These types of technologies help abstract the physical and create powerful logical layer which helps application and infrastructure resiliency.

Cloud computing can be a powerful tool. However, not all workloads are designed the same and certainly not all cloud models are built for the same purpose. Fortunately, cloud and replication technologies have come a long way, making it much easier to migrate massive workloads. Regardless, this does take up extra business cycles that could have been dedicated to more proactive projects. One of the best recommendations around the cloud is to plan twice and deploy once.

About the Author

Bill Kleyman is a veteran, enthusiastic technologist with experience in data center design, management and deployment. His architecture work includes virtualization and cloud deployments as well as business network design and implementation. Currently, Bill works as the National Director of Strategy and Innovation at MTM Technologies, a Stamford, CT based consulting firm.

Add Your Comments

  • (will not be published)