When the cloud became popular earlier this decade, conventional wisdom had it that the transition to an off-prem model would help enterprises become more agile, more competitive and reduce the cost of managing all their data.
The formula worked, but only to a degree. The cloud helped organizations become more competitive. However, organizations are belatedly realizing that the cloud does not necessarily reduce overall expenses.
For example, the cloud isn’t necessarily cheaper from a data storage perspective. The typical cloud customer pays the likes of Amazon, Google, and Microsoft a lot for their respective storage services – sometimes, disproportionately so. At the same time, the so-called egress cost, which is the expense incurred when you pull data out of the cloud, adds up in a hurry.
Even as recently as five years ago, the mantra in the enterprise world was cloud-first, with companies rapidly shifting application development and workload management to public clouds. Now, the pendulum is swinging in the other direction as organizations strive to have the right data at the right place at the right cost.
This doesn’t mean organizations are going to jettison the cloud and return to a data center infrastructure model, circa 1995. But as they eye current and future deployments, there’s going to be increased pressure to better manage their data, resources, and workloads with a closer eye on expense controls. A personal aside: I worked with a client that adopted a massive cloud-first strategy five years ago. Now they need to invest more in their on-prem infrastructure to balance out those costs. That used to be an anecdotal one-off; nowadays, it’s just one of many similar stories.
Managing Data Sprawl
While none of us have a crystal ball, it is not hard to see that there is a massive amount of macroeconomic uncertainty, especially in technology at the moment. But the signs are pointing to at least some kind of economic slowdown. Indeed, a recent Conference Board survey found that 98% of the CEOs surveyed were preparing for a recession. As we head into tighter financial times, cost optimization challenges posed by public cloud use can’t be ignored any longer.
As an industry, data sprawl has been a common topic for years. What are you backing up? How are you backing up? What’s the critical data you must back up? But that conversation was largely about data management. Now that conversation must focus on actual data costs.
If a recession does come to pass, there is no doubt that it will trigger cost cutting moves, particularly as it applies to IT infrastructure, whether that means on-premises infrastructure, public cloud infrastructure or hybrid cloud infrastructure.
Conventional wisdom suggests that once public cloud consumption attains a certain size, a customer’s cost savings begin to diminish. Along with the anecdotal evidence, there’s a great opinion piece by the venture capital firm, a16z, led by Andreessen Horowitz, that explains this from another perspective.
The authors make the argument that the biggest publicly traded Software-as-a-Service companies would potentially be able to unlock billions in market capitalization simply by shifting from the public cloud into a colocation model.
This more hybrid cloud approach allows certain infrastructure assets to be kept on premises and allows customers to consume only what is needed. For instance, Dropbox was able to save $75 million over two years after it moved most of their workloads from the public cloud to cheaper, custom-built infrastructure in colocation facilities that Dropbox leased and operated. Essentially, Dropbox found a way to better manage cloud resources while keeping a lid on cloud costs for those data and data movement resources to something that was more manageable and predictable. Significantly, the company’s gross margins soared to 67% – from 33% between 2015 and 2017, an increase that was primarily attributed to their “Infrastructure Optimization” initiative.
That article got a lot of deserved attention, but it underscored an extreme swing of the pendulum to make its point. In short, that means that many enterprises should not and likely will not make a wholesale move out of the cloud and move back to a data center infrastructure model. Rather, it will fall to the average organization to find the right balance that works for them.
If you believe that this is a long-term trend and not just a short-term blip for managing cost, the future is clear. It starts by recognizing the problem of data sprawl. Then it’s a question of deciding whether to manage on-premises or manage with a colocation provider. Most companies will continue to have a footprint in a data center and a footprint in the cloud. It’s just a matter of managing resources and workloads from a cost performance and business outcomes perspective in order to get the right balance for your business.
About the author: Greg Knieriemen is the Chief Technology Evangelist for the technical sales team at Hitachi Vantara.