I know, I said this would be a 3-part series! But as I got to writing and speaking with my peers and others about the state of things, a question kept rearing its head (in one form or another), “What about just moving to the public cloud?” So, I decided to weigh in on that line of thinking.
Public Cloud
In cloud adoption, most organizations are using it to some degree for productivity applications and email, and many more are writing applications to take advantage of the myriad of “aaS” options that can significantly streamline application development and innovation. But the story gets a bit murkier when it comes to adoption for more traditional workloads running on IaaS. I want to be clear; for this blog, I’m talking about native IaaS services, not a VMware or Nutanix deployment running in a Microsoft, AWS, or Google data center.
Looking at the history of public cloud IaaS adoption, it looked like the hyper-scale cloud providers would take over most of the workloads running in traditional data centers. The scalability, rapid time to deployment, and the end of managing costly and ever more complex on-prem environments were strong drivers for moving into the cloud. Early adopters found things weren’t as simple as abandoning their on-premises environments in favor of an “all-in” approach with the cloud. There were 1) security and compliance concerns to be worked through, 2) application and OS version compatibility issues, and 3) an entirely new operating model that required diligence to avoid unexpected cost overruns and run-away environments. Even in well-managed cloud environments, the cost associated with running steady-state IaaS workloads was often far more expensive than running in an on-premises data center, leading to the hybrid-cloud paradigm we see today.
As public cloud offerings and the operating models employed to manage them matured, most concerns around security, data sovereignty, and compliance have primarily been mitigated. When deployed and managed correctly some of the cost considerations have been (if not mitigated) justified for specific environments. When might an organization consider abandoning the on-premises data center altogether? Let’s explore.
Organizations that are “born in the cloud.” One of the most significant costs associated with the spin-up of an on-premises data center environment is the initial acquisition and spin-up - facilities, servers, networking, storage, etc. For new companies, this initial capital expenditure can be avoided by leveraging public cloud services from the very beginning. It also allows organizations to organically focus their IT efforts on application development and cloud management (leading to the most overused buzzword in our IT lexicon today, ‘DevOps’ – sorry, I had to throw that one in) rather than needing infrastructure admins with varying siloed specialties. Realistically, these sorts of organizations are few, with many IT shops being well established and a long history of IT operations that must be accounted for.
I mentioned earlier that steady-state applications have proven significantly less expensive to run in an on-premises data center, but what about volatile workloads? Workloads that require a significant amount of scaling up and down of resources to meet the ebbs and flows of demand can be far more effectively deployed and scaled in a public cloud environment where they are not limited by available infrastructure resources at the peaks and where they can be scaled down at low tide to avoid unnecessary cost. In scenarios where a significant portion of the infrastructure is deployed to support applications of this nature, efficiencies can be gained by running a limited footprint of steady-state applications in the public cloud along with the more dynamic workloads to an overall benefit. Making this shift, though, requires significant planning and retooling both in terms of technology and technologists, which in and of itself can be costly.
Organizations with tiny IT footprints can also benefit from shifting to native IaaS services. Small IT shops with a limited budget for equipment and skilled IT personnel can use the public cloud’s IaaS to eliminate the care and feeding of on-prem infrastructure. Or, as we’ve seen in many cases, the lack of care and feeding has led to a significant amount of technical debt, which now serves as an active detractor to business functionality. Moving to the public cloud can be a boon for these organizations.
The “hybrid cloud” notion is being adopted and championed industry-wide for a reason. For most businesses, there remains a reason to run public cloud appropriate workloads in the public cloud and steady-state applications, along with those that are data heavy or require super low latencies to remain in a local data center. The need for on-premises virtualization and orchestration capabilities isn’t going away for most, and the discussion flips back to which set of tooling is right for you.