The term “cloud” has stopped being useful as a description of a specific technology and business model. Everything, it seems, has some element of cloudiness. The definition of cloud versus on-premises has blurred.
It’s only been eight years since Gartner® defined the five attributes of cloud computing: scalable and elastic, service-based, shared, metered by use, uses internet technologies. Shortly after that, Forrester® defined the cloud as standard IT delivered via internet technologies on a pay per use, self-service model.
What we call on-premises is most often virtualized, dynamically provisioned infrastructure on a co-location hosting facility, programmable by software. Clouds now offer long-term, bare-metal, prepaid-for-the-year infrastructure, and private/dedicated infrastructure.
Organizations have learned that there is a place for cloud-hosted resources and a place for on-premises resources. The best analogy I have is this: there is a time when you want to buy a car and there is a time when you want to rent a car (or get a taxi). As Lew Moorman, president of Rackspace® told me many years ago, “the cloud is for everyone, but not for everything.”
It’s undeniable that every IT department is adopting the cloud, but it is also becoming increasingly clear that on-premises IT is not going away. Most companies will end up with a combination of the two. But how?
For the near future, there are mainly three broad ways to consume cloud by IT departments:
- SaaS – From SalesForce® to Netsuite® and Office 365®
- “Lift and Shift” – Where the architecture stays the same, and you only migrate the workloads to be hosted on a cloud
- “Cloud first,” which takes full advantage of cloud architecture and services. This model is only viable for net new projects and for those where it makes sense to invest in writing or re-writing apps from the ground up.
The reason I bring this up is because when it comes to monitoring, application architecture is more important than where things are hosted.
A standard three-tier architecture application like SharePoint® on AWS® needs to be monitored essentially the same way it is monitored on-premises, or in a co-location environment. Conversely, a cloud-architected application (service-oriented, dynamically provisioned, horizontally scaled, etc.) will require a different monitoring approach, whether it is hosted on a public or a private cloud.
The key point is that cloud is quickly becoming irrelevant as a term. No one says they have an electronic calculator or a digital computer anymore.
We need to start using more specific terms that are more meaningful and useful, such as cloud services, cloud architecture, or cloud hosting – not just cloud.