More enterprises are adopting a cloud-first approach in their IT deployment models, which is good news for the cloud industry but may not be in the best interests of the enterprise depending on how it is implemented.
According to Gartner, the shift to cloud computing will generate roughly $111 billion in enterprise spending this year, a level that will nearly double by 2020. That means the intervening years could see aggregate revenues of more than $1 trillion, a seismic shift that the company calls "one of the most disruptive forces of IT spending since the early days of the digital age." Part of this will be driven by the greater autonomy given to business managers to define and provision their own IT infrastructure, while part will come from formal policies from the front office mandating the cloud as the preferred solution for most general applications, reserving in-house infrastructure for the most critical of functions.
This trend is actually an offshoot of the zero-based budgeting that many organizations have adopted to reduce costs, says Forbes’ Joe McKendrick. The idea is to ditch the automatic renewal of line items to a system in which each and every expenditure must be justified every year to determine if it is, in fact, the most efficient and effective solution available. Naturally, the cloud wins over on-premises infrastructure in terms of cost nearly every time, so the only way to keep a particular app in house is to make the case for the added expense. As AWS’ Stephen Orban puts it, zero-based budgeting shifts the burden of proof from “why should we use the cloud?” to “why shouldn’t we use the cloud?”
But make no mistake, there are still some very powerful arguments against the cloud depending on what kind of application is at hand. Many legacy apps are not suited for the dynamic workloads and resource utilization patterns of the cloud, but cloud-native applications are not necessarily a slam-dunk either. According to Blue Coat Systems, virtually all business cloud apps fail to provide sufficient security and compliance controls, a problem that could become particularly acute given the rise of unauthorized “shadow” IT. At the same time, the company found that 95 percent of enterprise cloud apps are not even SOC 2-compliant, meaning they cannot be audited under existing SysTrust and WebTrust principles for things like availability, integrity and privacy.
And when it comes to apps that require low latency and high performance, it’s probably best to leave them in the data center, says Mike Vizard for Baseline. Organizations like NASDAQ, for instance, have found that trying to run high-speed trading apps across a wide area network is simply not feasible – unless you’re OK with losing potentially millions of dollars in the few seconds it takes to send orders to the exchange. This still doesn’t include many back-end processes, of course, but it points up the fact that when time, and money, is of the essence it may be best to maintain your own data infrastructure.
Of course, cloud-first is not the same as cloud-only. It just means that the cloud is no longer considered the dumping ground for apps and data that are clogging up legacy infrastructure. Rather, it becomes an equal partner in a rapidly evolving distributed, multi-layered data ecosystem.
The cloud provides unique advantages that cannot be found anywhere else, but you still have to take care when bringing it into the enterprise fold.