re: Why Infrastructure As A Service Is A Bad Deal
I think you make some good points about trying to ensure you aren't swindled by IaaS vendors like AWS, but I don't think you give enough thought to other costs and factors that make up the value proposition of going with an IaaS solution. Being an "as-a-service" offering, it provides a different (read: simplified) way of getting the job done. I agree that a lot of thought needs to be put into purchasing computing resources, but to write it off as a wholly bad deal seems unfair. As long as a company knows what it means to be an "as-a-service" offering, IaaS could be a perfect fit for their business needs.
As with most other "as-a-service" offerings the price is greater than what you pay to do it yourself because you're leaving some of the decisions that go into maintaining your infrastructure to someone else for a fee. Some simple examples include: Which server company should I buy from? What are the costs associated with bringing people in to install and/or maintain the servers? How will the resources be electrically powered? Where will they be stored? Who will physically secure them? IaaS takes care of these and more. The upfront management costs can serve as barriers to entry for smaller companies without big IT departments that are trying to do application development and analytics work. Not having to worry about all of these things to get a project moving can appeal to those smaller companies who are willing to pay more per month and avoid the initial costs.
Another key aspect of "as-a-service" offerings is that they focus on providing the service in near real time. When an organization wants to start a project that requires the acquisition of computing infrastructure, it can take months before development can begin due to the lengthy acquisition processes most companies and agencies must go through to get their resources. Many project ideas get thrown out because they're perceived as being too difficult to get underway. IaaS allows development to start in a matter of hours to days instead of months.
Finally, IaaS isn't permanent, which means that you don't have to buy into a long-term infrastructure to do a pilot project. If the pilot project turns into a more permanent engagement, then companies should definitely reconsider how they've setup their infrastructure to optimize their setup over in the long run, but investing in physical infrastructure in stead of IaaS can tun into a big loss if the pilot project isn't continued.
I'm not saying that AWS is doing IaaS the right way, but I am saying that IaaS is not a flawed concept. It could always be cheaper, and from your previous article, I agree with you that it should be.