There's lots of advice about Cloud technologies, yet some recent articles around AWS reminded me how our expectations of traditional hosting have changed. A business used to simply transition from one hosting provider to another via a simple "fork lift" for cost cutting or resiliency. After working with AWS, we’ve noticed an almost unanimous consensus that traditional datacenter moves aren’t ideal because of fundamental paradigm shifts in IT services[1]. One of my takeaways from visiting AWS headquarters was a paradigm shift for any Cloud enablement strategy, whether using Amazon technology or not. Here are 3 paradigm shifts that have changed the way technologists think about IT and Cloud:
- Pet versus Herd - the "pet versus herd" mentality is a crude term that reduces compute resources to a commodity, yet exposes emotional baggage. Technologists traditionally treated their systems like pets, including the humorous personification of computers, whereas Cloud treats systems like herds, including the expected losses and gains of components. This has little to do with resiliency and everything to do with perspective. A less caricatured comparison is seeing how the traditional datacenter design goal was to provide always-on services by keeping the component instances always-up, while the Cloud design expects components to be in various states of utility but the services themselves to never be interrupted. This paradigm shift has been epitomized by Netflix's Chaos Monkey, which tests the AWS mantra of “plan for failure and nothing will fail”.
- Infrastructure as Code - some have assigned the concepts of "infrastructure as code" to a new role called DevOps, but whatever you call this change, the fundamental shift is:
a) from infrastructure spec’d out for application needs
b) to the application building out the infrastructure it needs.
The difference between (a) and (b) may seem semantic but the implementation of this shift has caused debates to erupt between traditional developer and operation silos, such as the infamous nightmare of developer's "code running servers amok"[2]. There are various Cloud services that enable infrastructure components to be dynamically called by applications but the key shift happened when applications could build infrastructure via APIs.
- Don't repeat yourself - a key differentiator with Cloud versus traditional datacenter hosting is using services the Cloud provider offers. Some of these services are akin to SOA (Service Oriented Architecture) but a more modern concept is IaaS (Infrastructure as a Service) that keeps popping up in Cloud discussions, as well as automation. Automation always reeks of job insecurity in IT, but actually addresses long standing waste in operations where we think our infrastructure requires some unique solution but really boils down to reinventing the wheel. The shift is similar to developers switching to "don't repeat yourself" frameworks.
Because of these paradigm shifts, attempting a "fork lift" migration into a Cloud like AWS is not as simple (or pretty) as technologists might expect. An old school migration offers little to no advancement in technology and minimal ROI. Cloud enables a broad range of infrastructure services with APIs that change the way technologists design solutions. By shifting to the new paradigm, technologists embrace the latest advancements in agility, reclaim time for genuine intellectual progress, and lower IT costs. At ICF Interactive, we recognize that fundamental progress has been made by Cloud technologies and have committed our resources to expanding our Managed Services offerings that embrace these paradigm shifts.
References: