small glimpse of sky
Cost, security, performance and manageability are the four factors pushing analytics into its hybrid future PHOTO: Sharosh Rajasekher

Stay on-premises or move to the cloud? That’s a conversation that has been playing out for years as IT administrators who manage infrastructure and operations attempt to define the future of their organizations’ data centers. Though the conversation is ongoing, and the fate of the data center remains undecided, most signs now point to a hybrid future, one in which organizations leverage a blend of both cloud-based and on-premises technologies to deliver the optimal data center configuration.

Now, as the demands of analytics and data sciences increasingly exceed the capabilities of organizations’ aging on-premises IT systems, a similar conversation is heating up among CIOs, directors of analytics and chief data officers. Where does the future of analytics reside? Should we continue to expand the analytics processing power within our on-premises data centers? Should we offload some of our analytics applications and systems of record to the cloud?

Once again, the answer — and the future — will likely be found in the form of a hybrid approach. Let’s examine the issue in more detail, taking a closer look at the four key issues likely to drive analytics toward a hybrid future: cost, security, performance and manageability.

Not Cost Control — Cost Predictability

Pick an issue, any issue, and cost will be at the forefront of the discussion. Analytics is no exception. The hybrid future of analytics will be driven in large part by cost — specifically, by cost predictability. The nature of analytics and the divergent needs of multiple analytics projects within the same organization is such that traditional side-by-side cost comparisons are of little utility to analytics leaders. Simply comparing the capital expenses of an on-premises implementation to the cost-per-use cost of a cloud deployment fails to account for the variance that exists from one analytics project to the next.

For example, in one instance, an organization’s data science team might be tasked with managing experimental, innovation-oriented projects that come with enormous-but-temporary computing and storage requirements. Once the project ends, so does the need for all that storage and computing power. In such scenarios, organizations will likely want to use systems they can easily “shut down” when the project ends, therefore making a cloud approach more economical.

At the same time, that same data science team is likely also running mission-critical analytics projects, such as production reports, executive dashboards and customer behavior prediction models. Such uses require systems and infrastructure to be available 24/7 so that business analysts, managers and executives around the world can query data as needed. Renting anything on a 24/7 basis makes little economic sense.

Therefore, what analytics leaders need is not cost control so much as cost predictability — i.e., the knowledge that we’re likely to spend this amount for this set of initiatives and that amount for that set of initiatives. That’s something they’re most likely to get with a hybrid approach, one in which they can deliver round-the-clock availability of certain analytics systems while simultaneously delivering highly scalable, on-demand computing and storage capabilities when and where needed.

The Security Fight Continues

Marketingspeak being what it is, many cloud vendors are aggressively claiming to have won the war against hackers with strengthened security measures. Certainly, significant advances have been made. As any security expert will tell you, however, security is a never-ending battle and therefore remains a priority for most CIOs and CTOs, especially as the rate of cloud adoption continues to grow and the borderless ecosystem continues to expand. Among the many reasons organizations continue to invest in on-premises systems, security-related concerns, such as threats to data privacy and the risk of breaches, still weigh heavily.

As they ponder the future of their analytics investments and initiatives, IT leaders will be forced to balance fast, elastic delivery of high-powered, scalable analytics with the need to protect data and information from threats, both internal and external. Once again, the breadcrumb trail points to a hybrid future — one in which organizations keeps mission-critical analytics initiatives on-premises and inside the firewall, where they can control their own destinies with respect to security, while still maintaining the flexibility to move select initiatives to the cloud, in a secure environment maintained by an established and trusted partner.

Analytics Performance Matters

One constant in the ever-evolving analytics landscape is that performance matters. And when it comes to performance, there’s no denying that speed and scalability make your analytics more agile. In many ways, the lore of the cloud, both in general and with respect to analytics specifically, is that it offers the ability to achieve limitless scalability because it gives you access to an endless supply of computing power. The more scalability you have, the more agile you’ll become, thereby improving your team’s ability to handle analytics requests and respond to the needs of the business.

All of this is true — and if it were true without qualifiers, it would seem to point everything and everyone involved with analytics toward a cloud-centric future. But, of course, there are qualifiers. (Aren’t there always?) For starters, the aforementioned balance of cost and security means that organizations can’t really achieve the limitless performance that cloud computing seems to promise, at least not without limitless budgets and a willingness to set aside security concerns. In addition, just because they’re on-premises doesn’t mean the analytics systems on your side of the fence are getting any less innovative or any less powerful. Performance and scalability aren’t the sole domains of the cloud.

The takeaway? A hybrid future, of course — one in which both hosted and on-premises systems are capable of delivering the analytics performance an organization desires.

Is Any of This Manageable?

All the cost savings, security and performance scalability in the world won’t matter if the analytics systems they power aren’t truly manageable. At the end of the day, someone has to actually make all of this do what it’s supposed to do. For the people responsible for putting systems into production, manageability may be the single biggest concern. And if more than 20 years in the IT professional have taught me anything, it’s that manageability is not a black-and-white product feature. Manageability is subjective. It is perceived. It lives in the eye of the beholder.

One group of users might find a system remarkably easy to use, while another might find the exact same system to be kludgy and inefficient. This has been true as long as IT systems have been around, and I’m betting it will always be true. In other words, cloud computing will never “win” the manageability battle. Neither will on-premises systems. As a result, given the importance of perceived manageability to the analytics buying decision, it’s difficult to see either cloud-based or on-premises systems becoming the de facto standard for analytics. What’s not difficult to see? A hybrid future in which both remain in vogue.

Lead the Hybrid Drive

So what does all of this mean for analytics practitioners? Well, if the future is hybrid, then the true analytics leaders and visionaries will be those who drive their organizations toward that future by advocating and implementing a combination of on-premises and cloud-based systems. Leaders evolve with the times, and analytics is clearly evolving toward a hybrid future.