Are Enterprises Shifting Back to On-Premise Hardware?
The past five or ten years have been jam-packed with cloud computing hype. Indeed, the cloud is here to stay, without a doubt. But recent reports show analysts expect hardware sales for on-premise enterprise IT to tick up significantly.
High profile examples like Dropbox show that moving back to a more traditional data center can create efficiencies and free up cash flow. Is the enterprise data center – and by extension, colocation – about to put up a fight against the cloud?
Analysts Expect Hardware Sales to Increase
The growth rate of IT hardware sales has decreased in recent years, some of which is attributable to the siren song of the cloud. A new report from Morgan Stanley states that 2018 could reverse that trend, with double-digit earnings growth on the table for IT hardware manufacturers.
What might be causing this shift? The report calls out increased cash from recent tax changes, as well as the advantages inherent in the depreciation of hardware (which has been an issue when balancing the books on cloud computing, as described by our CFO Jason Graf in a recent blog post).
When a company buys new hardware, that equipment can be broken up in the Profit & Loss balance over the course of its anticipated lifespan, helping accountants boost EBITDA when compared to the monthly recurring expenses of cloud computing, which are operational rather than capital in nature. Buying hardware for some purposes could therefore be more attractive from a business standpoint.
Lower memory costs and a preparation for AI and the Internet of Things were also listed as reasons for more hardware sales in 2018. Keep in mind that major AI and IoT initiatives and research are being performed by enterprise cloud computing providers like Google and IBM, however, and that these hyperscale providers, alongside smaller cloud service providers like Green House Data, may actually be the purchasers of much of the hardware in question.
Dropbox Successful in Moving Away from the Cloud
Some have taken these signs as indicators of a larger shift towards enterprise organizations bringing their infrastructure back on premise in order to better design around their specific uses, gaining efficiencies alongside depreciation benefits.
One major example is Dropbox, who famously dropped the Amazon cloud in order to build its own infrastructure for its cloud storage offering. It took years, but the company has boosted its finances by migrating on-premise from AWS.
Of course, Dropbox likely couldn’t exist without having launched on AWS, allowing it to test, develop, and scale its platform over time without having to tinker with hardware and its associated operational costs. The key is scale — and of course the underlying product. At some point, the monthly costs of cloud, as well as the relative inflexibility of hyperscale platforms, becomes a drag compared to a custom-designed data center.
Not So Fast…the Cloud Still Has Its Place
No doubt Dropbox is still using pieces of public cloud infrastructure to supplement its own compute infrastructure. Indeed, the Morgan Stanley report does mention that much of the reason hardware sales slowed was that companies were getting their cloud strategy sorted out. Now that they have a better idea of what pieces to put in the cloud and which to keep on premise, they can begin spending on hardware again.
The RightScale State of the Cloud report shows that cloud spending continues to grow, with 38% of enterprises listing public cloud as top priority and 45% listing hybrid cloud as top priority.
Hybrid infrastructure, with some workloads running in cloud provider environments and others running in colocated or on-premise data centers, seem to be the mode that will last for some time. But there are some specific circumstances in which your product, application, or service may be much more suited to an on-premise deployment.
- You have cash to spend on hardware, which is often more costly up front than cloud
- You need to depreciate your IT costs over time
- Your workload is relatively stable
- You have specific performance or compliance requirements
- You need more visibility and control over security
On the other hand, if your workloads are variable, you need to scale, you have a limited budget, or you lack hardware expertise, then the cloud is a better starting point.
Ultimately there is no single best deployment model for enterprise IT infrastructure. Different workloads work better on different platforms. Bundling compute resources with software may make better financial sense with a specific vendor. A hybrid infrastructure might balance compliance with affordability and flexibility.
If you need help deciding how to architect a hybrid mix of IT, a provider like Green House Data is the ideal partner. With a wide range of experience in data center operations, managed services, and business software, we can help you decide where your workloads will operate with the best balance of affordability, performance, and security, even over the long-term.