While creating the equivalent of a physical data center in the cloud is simple, it is not economically advantageous. All the costs of running a data center still exist in the cloud and the per-unit price for things has to allow enough margin for cloud providers to make a profit. If you don’t change your operational model your costs will go up.
The thing that creates the opportunity to save money is changing how resources get allocated. In the cloud you can add resources as your load increases, but you can also remove resources that you're not using as your load decreases. This is a dynamic process that can be adjusted to react on a day-to-day, hour-to-hour or even minute-to-minute basis. When your environment adapts to its immediate needs, your costs go down because you stop paying for things that you don’t need. It’s like shutting the water off at a faucet when you’re done washing your hands. The cloud providers can offer this level of granularity because they have so many customers that overhead expenses are easily shared and the demand on equipment tends to level out as some customers increase their demands while others reduce it.
There are great advantages to being able to scale your computing environment up and down as needed, but it can’t happen by itself. Determining when to make adjustments and setting up a process for managing changes is an engineering problem that requires analysis and design. While automatic management is ideal, many companies don’t consider it a priority and only make changes manually based when somebody notices a problem. Often, overloaded systems get fixed by adding more resources, but underloaded systems are left running. This again can lead to costs being higher than necessary. When multiple people have privileges to manage resources in the environment, it’s not uncommon to find systems running in an environment that have been unused since a prior employee left months or even years before.
Clearly the idea of automatic processes to manage the scaling up and down of infrastructure has great appeal, creating a process that functions effectively is not always straightforward. Automatic processes require ongoing monitoring and fine tuning. When running well, such processes can be enormously beneficial, but poorly tuned processes can lead to problems. Good automation needs to be able to ensure that an environment isn’t over-provisioned to save costs, but it also must ensure that it can react in a nimble way as the load goes up because excessive lag in performance can have negative impacts on customer satisfaction (and therefore revenue). Good automation needs to be set up to balance the needs and find that “just right” goldilocks zone of optimal operation.
Hidden Costs from Poorly Managed Cloud Resource
As a business grows, operational costs are naturally expected to increase. This is as true with cloud computing as with any other ongoing expense. However, when a cloud environment isn’t properly managed, it can easily be the case that a company ends up paying for more resources than it currently needs. Since the costs tend to increase in small increments over time, the chances are high that the wasted spending will go unnoticed for extended periods of time. In extreme cases companies may be overspending by tens of thousands of dollars per month. (This is not a hypothetical, iuvo has identified cases like this on more than one occasion.)
