Cloud repatriation may be a sensitive and polarizing topic for some, but it really should be just another architectural option.
When I say that some workloads should return to the data center, I am not pushing back on cloud computing. This is about merely finding more optimized platforms for certain types of applications and data sets based on changes on technology and market. The objective is to bring more value back to the business.
To determine when repatriation makes sense, we need to think about basic concepts that architects should be considering anyway as they look for opportunities to provide better cost-optimized platforms for some workloads.
Look at the monthly costs and values of each platform. This is the primary reason we either stay put on the cloud or move back to the enterprise data center. Typically the workload has already been on the cloud for some time, so we have a good understanding of the costs, talent needed, and other less-quantifiable benefits of cloud, such as agility and scalability. You would think that these are relatively easy calculations to make, but it becomes complex quickly. Some benefits are often overlooked and architects make mistakes that cost the business millions.
All costs and benefits of being on premises should be considered, including the cost of the humans needed to maintain the platforms (actual hardware and software), data center space (own or rent), depreciation, insurance, power, physical security, compliance, backup and recovery, water, and dozens of other items that may be specific to your enterprise. Also consider the true value of agility and scalability that will likely be lost or reduced if the workloads return to your own data center.
Make sure to do the same for the cloud side of things: the cost of humans to maintain the cloud-based platforms, ingress or egress costs (moving data in or out of a public cloud), security systems, backup and recovery systems, etc. Also factor in the likely price increases or decreases, now and into the future. Most importantly, is that workload leveraging native cloud services, such as AI and data analytics, that are not easy to replicate on premises anymore? Most investments and innovations now occur on cloud platforms.
But that’s not all. You need to consider the costs and risks of repatriation. If some refactoring was done to leverage cloud-native features when the workload was migrated, those will have to be dealt with. Also, the workload will have to be retested and tuned on the on-premises platform. Most people don’t understand just how complicated and costly that process is until they’ve done it.
Also, most people will have some office politics to deal with. You’ve moved a workload from on premises to a public cloud and now back to on premises. Can you explain why in a way that will keep you employed?
You need to understand the costs and benefits of each platform (cloud and not cloud) and model them for least 10 years. If they are about even, the tie goes to the cloud. Risk and repatriation often costs more than we expect, and likely the cloud platform will have some additional benefits because the R&D investment in cloud far exceeds that of traditional systems.
There needs to be a substantial benefit to moving a cloud workload back to the enterprise data center or even to a managed service provider or colocation provider to justify repatriation. The workloads that are good candidates for repatriation normally have one or more of the following characteristics:
- Not tightly coupled to public cloud services. Workloads using AI, data analytics, governance, operations, etc., are generally better on cloud.
- No special security or compliance requirements. This is related to the previous point. Many believe that on-premises systems would be a haven for security and compliance, but in most cases the opposite is true. Security is typically better on public clouds.
- Generic use of many CPU or storage resources. If you use a petabyte or more of storage and don’t use any special cloud-based features, on-premises storage may be way more cost effective given the drop in physical HDD (hard disk drive) pricing in the past 10 years. The same goes for compute, including CPU type (and other processors) and memory types and sizes. The prices of on-premises servers have dropped in some cases as well; however, the price of processors is more volatile.
- Less important to the business. This is a strange metric. The more important the workload is to the business, the less benefit you get from repatriation. This is due largely to the risk of outages in the repatriation and testing processes, and impact on revenue and reputation if the workload does not initially work as well on premises.
Nothing easy about this, as you can see. But it’s the job of IT and architects, specifically, to find the platforms that bring the most value back to the business. This problem is no different.