Is the Infrastructure as a Utility Dream Still Alive?
While we may wish cloud storage was as simple as turning on a water tap, the truth is the service is much more complicated… and costly.
In a recent article, Philippe Fosse discusses the cloud as a utility, claiming that cloud may reach a stage where it could be turned on and off like water at the tap. This analogy, while perhaps helpful in the early days of cloud, fails to register certain key complexities of the service. Furthermore, its deceptive simplicity may disguise the costly consequences of improper cloud usage.
A better analogy would be that cloud service is like an endless supply of Lego blocks. The scale of any project can be adjusted by adding on or taking away blocks. But the process isn’t simple, and each adjustment requires carefully refactoring the project as a whole. Let’s take a closer look at these two analogies and how they relate to our understanding of the cloud:
One of the driving notions behind the utility analogy is that the cloud can cut out wasteful spending by delivering only the storage capacity needed at any one time. As Fosse writes, “If you want more of it, the service provider can turn the tap on; you need less of it, they can turn it down.” By adjusting storage capacity over time, cloud could minimize costs while maximizing performance. Such functionality may stand as a lofty goal for cloud, but for now, such a goal is not only impractical: it’s deceiving.
As the Wall Street Journal has reported, cloud services often lead to just as much wasteful spending as traditional infrastructure, if not more. While ideally the cloud would only deliver the amount of storage necessary, in reality companies tend to over-order storage and keep the proverbial tap on longer than necessary.
This is largely due to a failure of knowledge: unlike most utility bills, people don’t know what they should be paying for storage and often err on the side of overspending. Without such knowledge, companies will continue to overspend, and broad claims of the cloud’s ease of use will not remedy the situation.
In the same article, Fosse discusses the cloud as a “bespoke service” that will automatically adjust to each user’s workload. The user won’t even have to turn on the tap — the tap will turn itself on based on usage analytics. But such a service depends on rigorous performance monitoring, which is actually more difficult on cloud.
There are several reasons for this. With traditional storage infrastructure, all data is in the hands of the individual company. Now, however, the cloud service provider owns much of the performance information and can choose whether or not to share this information. Additionally, if the cloud service provider is a relatively new company, it likely will not have adequate performance management systems in place.
Finally, regardless of cloud service provider, a company will now have to analyze data from two sources: the pre-existing storage infrastructure and the cloud. This necessarily makes performance monitoring more difficult, and can increase the odds of data mismanagement.
None of the above should imply that cloud should not be integrated into a company’s storage system. But the cloud should not be thought of as an easy alternative to traditional storage. Nor should it considered a hands-off, prepackaged service.
To get the most out of cloud, companies need architects to plan and monitor the service. As of now, cloud has only really improved computing and storage costs for companies who can properly manage their IT infrastructure. Before investing in cloud, companies should invest in skilled IT professionals and software solutions. After all, you wouldn’t expect your Lego city to build itself.