I was invited to speak at Interface 2024 – Seattle on the official topic of, “Navigating the Return: The Shift from Hyperscales to Private Cloud” as a guest of one of the sponsors of the event Wowrack. You can check out the video which is posted on YouTube here. During the talk, I explored some important considerations involved in migrating from public cloud to on-premise environments.
This talk speaks directly to firms that made the move to one of the hyperscalers with workloads but haven’t been able to capitalize on the most compelling trait of the cloud which is elasticity. So migrating from public cloud to on-premise is really becoming a frequent topic in my conversations.
Here’s a synopsis of the talk:
Why is This Happening
- Client bills are huge
- Externalizing responsibilities to the cloud only offload hardware responsibilities
- Clients unable to use the auto scaling of the cloud
- Clients that have an economy of scale that rivals some cloud providers
- Clients spending more in OpEx than initially spent on OpEx and CapEx previously
- Unusually high pressure to adopt cloud for the appearances of modernity
- desire to externalize responsibilities
- Overworked, under-funded, and understaffed IT personnel
- Cloud doesn’t allow for “slop” in extra resource provisioning.
Benefits of Repatriation
- It allows organizations to leverage their own economy of scale
- Predictability of billing
- Reminds us that security and scaling were always the responsibility of the enterprise and not the hyperscaler
- Allows for the correction of lifted and shifted workloads that weren’t compatible in the cloud
Challenges for Repatriation
- Depending on how much you cloud enabled your application, you may have to “undo” some cloud aspects (like using PaaS)
- If you re-hosted (lifted and shifted virtual instances into the cloud) you may be in luck and this may be easier than other migration
- If you re-platformed (lifted and optimized) this will require more work such as moving from RDS back to your own mySQL database
- Repurchased application (moved from self hosting to SaaS) this requires very little work
- If you refactored / re-architected, then your path is more complex and difficult and may warrant optimizing that workload where it stands and not repatriating that workload
- determining whether applications should retired or left alone
Strategies for a Successful Transition
- Be realistic, refactoring and re-architecting are not simple matters
- Get an external assessment of what’s involved and what it may cost. Some bare metal or private cloud providers would be willing to invest a lot of pre-sales engineering to make it happen
- Find a specilist that looks beyoind the hardware and understands PaaS and your application stack.
- Consider not moving everything, pick areas where meaningful change is possible to save cost
- Consider not moving and optimizing everything in place. Be wary of “AI only” solutions to fix everything.
Again, if you’d like to check out the talk, I’ve linked it here. We have some additional resources available on this topic including:
– Data Center Migration Project Plan (more about data center to data center rather than public cloud to on premise)