Companies’ cloud strategies rely on two main points:
- A reduction in IT headcount. Hosting services in its broadest term means less administration by IT operatives employed by the company. This obviously saves money by way of lay-offs, but it also means that existing staff will not be required to know the intricacies of that system or service; it’s someone else’s problem. This stagnates people’s IT development and pushes the requirement to invest in employees away from a company.
- Increased Availability of that service. There is a assumption that on-premise is a lot-less stable than cloud services, therefore a move to cloud means more productivity because there’s less likelihood of the service being unavailable.
Personally, I am still very cautious to knee-jerk everything out to a service on the Internet. On-prem means total control over that service, it means that IT operatives remain skilled in that system, and it means that you can employ a very flexible strategy if you want. Sure, cloud services doesn’t necessarily mean all responsibility is absolved – for example Azure replaces your VM platform but still requires OS config and management by the “home team” – but I do not like being bounced back on training requests for my team because it’s a managed service.
Equally, third-party support is also used as an excuse to de-skill the department. Sure, no-one knows a system better than a company or vendor that deal with a particular system 24/7, but in order to properly analyse issues, home-team IT operatives need some skills in a system to diagnose problems in the infrastructure. I have had this issues countless times, from being locked-out of the core switches because it’s under 3rd party support, to having almost zero DB skills because a third party supports a particularly complex Oracle DBMS.
This kind of strategy is cyclical; first we had a single computer with all its resources local, then mainframes, where all resources were remote. Then the processing and storage resources went out to the clients in a traditional client/server setup, and now all the power has gone back centrally in what I call a cloudframe topology. It’s the circle of IT life.