Another bubble to burst?
Depending on where you are in your journey with XaaS, cloud and utility based consumption models, speed comes at a cost. You have to make a decision about which levers to pull and when to pull those levers. Myself, I am in the process of moving my home environment into a cloud provider. While my applications and workloads aren’t considered production or mission critical since I have no articulated revenue stream associated with them, they are still important to me and importance is only purely relevant to its audience.
While looking at my vm’s, networking, storage, etc running on my BYOC (bring your own crap) environment at my house, I quickly realized that the complexity of moving everything over, whether it be right away or over time, is something that companies and organizations seem to overlook or forget. So you’ve decided you’re going to move part or all of your workload to a cloud provider? Great! What happens when a CFO asks you the question of “how much is the migration going to cost and how long will it take?”
Let me reveal the magic behind the curtain in case you haven’t quite figured it out. There is no simple, easy or cheap way of simply lifting and shifting applications from on premises to a hosted or cloud environment. There is no button or magic wand you can wave to make this happen, outside of yelling beetle juice three times in the mirror!
First, not all applications are going to the cloud. Some applications are simply unable to be moved: think an application that has been written in house by developers that are either no longer working at the company or an application that is written in a language that is no longer supported. Second, some applications will simply be decommissioned and of course will not move. Finally, for those apps you choose to move, you can of course replatform the application (typically legacy applications) and then rearchitect it for services on your cloud providers environment. For something like AWS, this might mean looking at services like EC2, Cloudfront, Route53, DynamoDB and Elasticache. For others like Google Compute Engine, this might mean BigQuery.
Regardless of what cloud provider you choose please adhere to following when thinking about migrating on-premise workloads to the cloud:
- Get help early. This could be in the form of a partner, solution architect or system engineer.
- Go all-in. In my case, my workload use case is small so I am going from all hosted in my house to all hosted in a cloud. The economics, management and ease of use for my situation easily justify this but please run a TCO on your own environment.
- Include security and compliance requirements
- Ruthlessly automate. Whether this means using new tools, scripting or both, automate, automate, automate!
- Establish application performance baselines
- Optimize both on and off prem costs. If you decide to move 80% of your workloads off prem and still own or pay for a data center, you are paying for 100% of the data center usage while only consuming 20% of it.
What tips and tricks have you learned during your migration?
Have you moved some or all of your workloads off prem?