Writers like making big, powerful statements in their blog post titles. We do it, too. Adding numbers works particularly well.
Recently, a post titled “Why There’s No Compelling Use Case for Hybrid Cloud” went up on TechRepublic. In the post, the writer illustrates that the few scenarios where it would make sense to use a Hybrid Cloud are not realistic, and there is, therefore, no real use case for a Hybrid Cloud at the enterprise.
The main arguments in the article are around cloud-bursting and the challenge of moving legacy applications from a VMware datacenter to a Private Cloud.
Regarding these two use cases, I have to agree. Cloud-bursting, the ability to scale an application from the private to the public cloud to optimize resource usage, is undoubtedly cool but very rare. Not a typical scenario for enterprises.
Migrating a monolithic application that relies heavily on VMware from a datacenter to a cloud is also a herculean task. So once again, not a valid reason for an enterprise to adopt a hybrid cloud.
The final nail in the Hybrid Cloud’s coffin, according to the post, is that a centralized management model where developers write to a single API and a cloud management platform figures out where to put the workloads - is too difficult for most enterprises to implement.
While the article makes some good points, I think it misses a larger point. Mainly, Hybrid Cloud is not about individual applications.
Hybrid Cloud is IT’s Problem, Not The Developer’s
The reality is that most enterprises find themselves in a de facto Hybrid Cloud situation whether they like it or not. Most enterprises have a massive VMware deployment that’s been running for years, and maybe one Business Unit has an OpenStack project going on and another department is leveraging AWS and Azure for their applications.
Each solution, private cloud, datacenter and public cloud, solves a different problem. Enterprises are using all the tools they can to create workflows that let them move fast while remaining compliant, secure and within budget.
The Hybrid Cloud typically manifests not at the provisioning level, with an incredible cloud-bursting superpowered app. It happens at the corporate IT level, where an administrator needs to manage, monitor and provide services for private clouds, public clouds and datacenter.
Maybe a single application doesn’t need to span multiple clouds, but the organization, the budgets, and the corporate policies do.
Another point we need to address is the argument that the centralized management model, where the CMP decides where workloads go, isn’t feasible. I agree with this one.
The CMP shouldn’t magically decide if an application goes on AWS, Azure or OpenStack. Workload placement requires a lot of context, so for the CMP to make a perfect decision, it needs perfect data - which would require the user to supply an endless amount of information. Instead, the CMP provides the user with the relevant information, and depending on the workflow, the admin or user makes a decision. While having a report that says “if you put this on GCP, it would be cheaper” is nice, there’s usually a lot more that goes into deciding where each Business Unit and departments will do their provisioning.
Having a standard workflow for consuming cloud resources lets enterprise IT defend one gate instead of a hundred, and it’s the only way to responsibly provide self-service to employees while making Security and Finance happy.
The approach more and more enterprises are adopting is centralizing policy enforcement while decentralizing access. Match users with the provisioning method that makes them most productive, like an API, a power console or a simplified storefront, but enforce policies centrally.
In short, yes, there are compelling cases of Hybrid Cloud, and it very much exists in plenty of enterprises.