BrainBlog for RackN by Jason Bloomberg
Many enterprises find that the cloud doesn’t live up to the hype. As a result, they’re moving workloads back to self-managed infrastructure on premises – a trend we call repatriation.
Cost savings is the most common motivation for such a move. Cloud bills are generally usage-based, so the more you use, the more it costs. At some point, managing your own infrastructure will cost less.
Other repatriation motivations include data sovereignty issues, latency concerns, and specialized application needs. Regardless of the reason why, organizations must now contend with managing their own infrastructure once again, long after they moved much of it to the cloud.
Nobody wants to return to the bad old days of managing on-premises infrastructure. Building out and deploying servers one at a time was expensive, time-consuming, and often resulted in brittle deployments.
The good news: such do-it-yourself (DIY) build-outs are a thing of the past. Today’s operators must bring cloud best practices in-house so they can effectively self-manage their infrastructure.
From Pets to Cattle, not Vice-Versa
Cloud aficionados are fond of likening the move to the cloud as ‘cattle, not pets.’
Before the cloud, operators set up and managed servers individually, even going so far as to give them names – just like pets.
But in the cloud, operators leverage automation to manage fleets of servers as a group. If one has a problem, they replace it and reprovision it – much like anonymous cattle. The herd is the focus rather than individuals.
The ‘cattle not pets’ principle has come to represent the way cloud providers manage their gear, leveraging automation and configuration-based deployments they refer to as infrastructure as code (IaC). IaC provides the processes, artifacts, and scripts that configure and reprovision servers and any other cloud-based assets.
When organizations repatriate their cloud-based workloads, thus provisioning infrastructure to support them, they may fall into the trap of moving from cattle back to pets. However, going back to the old days of manual provisioning and configuration would be a costly mistake.
Instead, such enterprises should take a page out of the cloud playbook, bringing automated IaC processes and configuration-based approaches to their new self-managed infrastructure.
Are We Recommending Private Clouds?
Private clouds have been around since the early days of cloud computing and have evolved apace. The idea is to set up one or more sites as though they were public clouds, only keep them under private control – by enterprise operators directly, or more commonly with the help of a colocation provider.
Does following cloud-based practices when repatriating workloads from the cloud to self-managed sites mean setting up those data centers as private clouds?
The answer: perhaps, but not necessarily. Over the years, private clouds have delivered mixed results, largely because vendors and service providers have treated them as black boxes. Hire this colocation provider or buy that software and presto! You have a private cloud in a box.
Such turnkey private clouds mainly consist of virtualization-centric servers and so rarely meet the diverse business needs of today’s enterprises. Every organization has business priorities, technical debt challenges, and IT strategies that virtualization cannot address.
Soon they end up once again with pets rather than cattle: DIY processes, policies, and configurations that may meet short term needs, but forego the scalability and velocity benefits that made the cloud attractive in the first place.
Self-Managed vs. DIY Infrastructure: Stick with Cattle
Organizations that wish to repatriate cloud workloads thus face an apparent dilemma. Do they move to a private cloud, even though its black box aspects may not be a good fit for their needs? Or do they return to the DIY infrastructure of old, thus losing the benefits of the cloud?
The good news: there is a third option. Self-managed infrastructure can still take advantage of cloud-centric processes and policies by leveraging IaC software from vendors like RackN. RackN provides configurable processes for provisioning infrastructure from the bare metal on up to the platforms and applications.
Starting with bare metal – hardware with no software installed – is the key. Bare metal allows for the installation and configuration of everything from the firmware to the operating systems to the hypervisors and execution environments. Bare metal thus provides an automatic foundation ranging from hardware to fully operational data center.
Unlike the DIY approach where operators figure out all the installations and configurations, with RackN they simply select the options they require, and the software handles the provisioning and configuration automatically.
The result is a self-managed data center configured just the way the organization wants, without having to build or configure custom automation. It’s the best of both worlds: control as well as standardization.
The Intellyx Take
It’s important for operators to think about self-managed infrastructure as a platform so that they avoid returning to the DIY approach of old. Instead gain the ‘cattle not pets’ advantages of the cloud – without the cloud vendors.
There’s more to this story, which we’ll cover in the next three articles in this series. Intellyx Principal Analyst Eric Newcomer will go more in depth in defining the cloud behaviors that repatriation should preserve. Then I will return to dispel myths around bare metal. Eric will wrap up the series to discuss building future-proof infrastructure with RackN.
In the meantime, feel free to plan your cloud workload repatriation initiative without fear that it won’t meet your specific business needs – while preserving what’s best about the cloud.
Copyright © Intellyx LLC. RackN is an Intellyx client. Intellyx retains final editorial control of this article. No AI was used to write this article.
