In-Depth
Cloud Adoption Isn't All or Nothing
The cloud might be the future, but there's no rush -- take the time to do it right.
You're a smart systems administrator who can see the writing on the wall: The cloud is the future. On-premises clouds might be part of the mix, but somewhere in there everyone is going to have to incorporate workloads that exist either on a services provider cloud, or on the cloud provided by the big four public cloud providers. How do you get there from here?
The initial assertion that the cloud is the future may strike some as bold. This is especially true of individuals and organizations that have yet to embrace the cloud. In reality, however, "no cloud" policies are increasingly rare. By 2020, analyst house Gartner Inc. predicts they'll effectively be nonexistent, and this is backed up by virtually all other analysts and polling on the subject that is publicly available.
Gartner also says that current public cloud spending is rather high ($219.6 billion), with predictions that the amount recorded in 2016 will nearly double ($411.4B USD) by 2020. While this should be taken with a grain of salt -- Gartner's predictions aren't exactly stellar when it comes to market size -- the actual numbers don't matter so much. What matters is that both analyst predictions and real-world evidence show continued strong growth, with no end in sight.
Where the statement "the cloud is the future" gets hackles up is that a decade of tech marketing has left people with completely different ideas of what "the cloud" means. For some, mentioning the cloud conjures images of Infrastructure-as-a-Service (IaaS) solutions like Amazon Web Services (AWS). For others, the cloud means Software-as-a-Service (SaaS) solutions like Salesforce.com.
We all use cloud services -- personally, if not professionally. Even if your worksite is inside of a hardened bunker in the middle of the high arctic, with no external network connectivity, somewhere in your life you are using an instant messaging application, Facebook, Twitter, or even just reading news articles online. Behind all of that, today, there is a cloud.
How we perceive cloud services determines to a large extent how we react to the idea of moving organizational IT resources to the cloud, and whether or not we realize we've already begun to do so.
Phasing In
An important discussion point when planning your journey to the cloud is the difference between a bulk move and slowly phasing cloud solutions in. Today, the bulk of virtualized organizations run VMware's virtualization solutions. Most do so without any self-service portals or IT automation solutions. In their quest to become more efficient, however, organizations are slowly adopting on-premises cloud technologies.
VMware's vCloud Suite is a popular choice for the creation of an on-premises cloud, in large part because it can be layered over the top of your existing VMware installation. Microsoft has its own offerings that layer atop its Hyper-V virtualization platform, and those invested in the open source world have OpenStack.
To varying degrees, each of these solutions offers customers with existing virtualization deployments a means to gradually ease into the cloud. Automation and self-service are simply added to what already exists and, voila: a wild cloud appears.
An on-premises cloud doesn't mean that the transition will always be gradual. If you purchase an on-premises cloud appliance and migrate all your workloads, that's a pretty blatant bulk movement of services.
This path to becoming cloud-enabled is less traumatic than a bulk migration to AWS, which is what most people think of when told to "adopt the cloud." Somewhere in the middle is the adoption of hybrid cloud solutions involving services provider clouds, (for example, the myriad vCloud Director-based VMware services providers,) or fully managed on-premises-compatible offerings like Microsoft's Azure Stack, or VMware Cloud on AWS.
Hybrid Clouds
Phasing cloud management software over the top of your existing virtualization infrastructure is the path of least resistance, but it only gets you so far. One of the touted benefits of cloud computing comes from the ability to outsource the underlying infrastructure entirely.
For most organizations this starts with the elimination of their disaster recovery (DR) site, and the adoption of a services provider or major public cloud provider as the new backup and DR destination. The right services provider can make all the difference for organizations cautiously adopting cloud computing, while a poor one can poison an organization against the cloud for years.
The major public cloud providers don't really have a great reputation for support or customer service, especially if you're a smaller organization. They stand up a fantastic cloud with a great UI and a top-notch API, but getting it running is left to the customer.
Smaller services providers, on the other hand, frequently specialize in meticulous support tailored to help organizations with no cloud experience take those first few cautious steps. Most smaller services providers put the time in to help customers configure their solutions appropriately, and develop testing and automation regimens that not only ensure solutions like DR to the cloud are working properly, they ultimately help an organization's IT team out by freeing them from mundane day-to-day tasks.
The helping hand of a managed services provider comes at a price. In terms of raw numbers, the major public clouds are always going to be less expensive. These hyperscalers operate at scale no other providers can even approach, and they can grind efficiencies out of their supply chain and operational management that the rest of us can only dream of.
The counterpoint to this is that most organizations aren't ready to play in the hyperscaler sandbox. If an organization was prepared to burn down their entire IT plant and rebuild it from scratch, learning entirely new skills, then the major public cloud providers are a great plan.
Most organizations don't have that luxury. They have to keep what they have running while they migrate to the cloud. A path that connects the two that doesn't involve jettisoning a career's worth of knowledge and experience is an attractive bonus.
For this reason, services providers clouds -- especially those offering hybrid solutions that build on an organization's existing virtualization platform -- remain popular.
Technical Debt
On-premises workload management differs from hosted workload management. When you own your own gear, standing up a workload and leaving it running 24x7 is the norm. Organizations design their datacenters to handle the peak amount of workloads, so there's little incentive to turn off idle workloads. This is a terrible way to engage with hosted cloud services.
As you're charged by the hour for workloads that are active, cost minimization on hosted clouds requires that workloads be active as little as possible. This is best accomplished through the use of composable workloads. At its most basic, composability has a workload's networking made dynamic and automated so that when a workload is moved from cloud to cloud it remains accessible.
A more complete version of composability separates the data from the application and the operating system environment (OSE). The OSE and application's configuration are laid out in a configuration file, and they can be instantiated as required, from scratch. In this case only the workload's data needs to be moved from place to place, not the entire workload. This makes standing the workload up and tearing it down as needed significantly easier.
Most organizations are nowhere near composability. Workloads with static IP addresses are common, and the use of configuration management tools like Puppet, Chef, Ansible and Saltstack remains rare. Adoption of more advanced infrastructure automation solutions like Terraform is still in infancy.
It's this technical debt that is the real barrier to simply packing up your IT and moving it up to a major public cloud provider. A VMware administrator can learn the AWS interface if he has to. But it's a long road from pets to cattle.
The cloud might be the future, but there's no rush. Take the time to do it right. Automate what you can. Build a private cloud and evolve toward hybrid capabilities. Embrace services providers and rely on their experience. The major public clouds will still be there when you're ready.
In the meantime, one application at a time, your organization is probably already embracing the cloud. From Salesforce.com to Office 365, workloads that once were fully managed by an organization's IT are now consumed as a service.
Cloud adoption isn't all or nothing. Don't let anyone tell you otherwise.
About the Author
Trevor Pott is a full-time nerd from Edmonton, Alberta, Canada. He splits his time between systems administration, technology writing, and consulting. As a consultant he helps Silicon Valley startups better understand systems administrators and how to sell to them.