Do Cloud Computing Platforms Require A New Way of Thinking About Infrastructure?
Infrastructure-as-a-service (IaaS), the ability to provision and leverage resources on cloud computing platforms, has very much arrived as a practice. It is no longer simply a fringe-IT trend or part of a future projection: it is actually the foundation for a good number of the services that we use every day.
With very small amounts of upfront money, any business can spin up virtual server instances running any of the major enterprise operating systems (e.g. Windows Server or Red Hat Enterprise Linux). Billing is usually based on operating time, and virtually unlimited storage is also available but charged based on consumption.
Where does the server reside? The matter is irrelevant, and due to the way IP networks work, the layers at which devices communicate with each other do not depend on the underlying infrastructure: a virtual machine is just as real as a “real” server when it comes to web traffic. A clever researcher reached the estimate that Amazon Web Services EC2 is using nearly half a million physical machines, but this hardware infrastructure is able to provide virtual instances to the entire world.
Of course, some would argue that we need to employ some novel thinking when discussing virtual instances. The CTO of Amazon, Werner Wogels, has referred to EC2 instances as “not a server” but “a building block.” The explanation for this, according to one cloud expert, is that persistence of the virtual machine is not necessary – it can come into being when needed. It is also suggested that EC2 owes much to service-oriented-architecture (SOA), the design approach that was instrumental as Amazon scaled its operations and eventually ended up with a cloud infrastructure that it was able to lease to others.
However, with a skilled consulting partner, a business need not know the full details of how to best operate on cloud infrastructure in order to get the most out of it.