Measuring big data might sound like a fool’s errand, but those in IT know otherwise. As the capabilities of modern data centers grow exponentially, metrics and key performance indicators (KPIs) have become increasingly important to understanding just what all the data means. And the most important metric of all is your bottom line.
It’s not a matter of just measuring, but managing. Providing SaaS products can be lucrative, but as with any business, there are pitfalls to be aware of. Let your data center develop too “organically” and you’ll be bleeding bitcoins out your ears. If, however, you know where to check costs and add structure, you’ll be on the way to a successful business.
Of course, there are a number of hazards associated with connecting your operations to the Internet—six percent of all computers suffer some kind of data loss annually, for example—but the benefits of a modern, connected business are too great to forgo. If you make sure to monitor your data center actively, you can ensure not only cybersecurity but cost efficiency as well.
Total cost of ownership (TCO) isn’t a new concept, nor does it need a high-level definition. TCO is just what it sounds like, but dive a little deeper and you’ll find some details to consider. If quoting the exact cost to run your operation was easy, there wouldn’t be books written about it.
Running a quick TCO calculation means taking into account two types of costs: direct and indirect.
- Direct costs come from assets, such as company cars, printers, power, network connectivity and payroll.
- Indirect costs are the costs indirectly linked to a final objective—for example, rent for office space, productivity lost when people take PTO and legal fees.
If you can accurately map out and sum up your direct and indirect costs, you’ve determined your TCO. Just getting that far is a great start. Now, how do you identify where costs are too high so you can begin to run your data center more efficiently and more profitably?
TCO in the Real World: IT Brand Pulse
IT Brand Pulse (ITBP) published a case study in 2015 in which it investigated the cost to own and operate several components of a mass-storage array. It took into account the purchase price of multiple hypothetical hardware components for the data center as well as secondary costs, including spare parts, training and annual support fees.
After explaining the method used in the report, ITBP explored the storage options available. It weighed the performance available from different solutions including Amazon’s cloud-based storage service, multiple disk arrays and a server-based software-defined solution.
By comparing six different storage solutions, ITBP demonstrated which of the five is the most cost effective. The Suse product it selected turned out to be more than 50% cheaper than Amazon’s cloud-based solution over time, even though Amazon touts the savings of offering one-cent-per-month storage. It pays to do your homework.
TCO in the Real World: Ericsson Hyperscale
A 2016 Case study by Ericsson demonstrated tangible savings from the company’s Hyperscale Datacenter 8000 system, categorized as capital expenditure (capex) and operational expenditure (opex). By reducing both costs, Ericsson’s system demonstrated a 138% return on investment (ROI) for its client—that’s a good way to keep customers loyal.
Taking a closer look at how the company achieved these savings reveals ways in which data centers can reduce TCO without hindering capability. The Ericsson system used less power and required fewer people to administrate than previous solutions. It featured hardware architecture that was both compact, minimizing space, and easily replaceable, minimizing downtime.
The new solution also delivered greater capabilities. New features encouraged customers to invest and invited confidence. For Ericsson’s Hyperscale system, these features included “infrastructure on tap,” which is billed as an “AWS-like” method of managing assets that customers are familiar with.
Additional savings came from the system’s low deployment cost and thoughtful low-maintenance architecture. Ericsson designed a proprietary racking system that maximizes space and allows for quick access during maintenance. The result was a savings of up to 74% in deployment costs relative to competitors.
Countering the physical savings, the system’s software was extremely optimized, allowing it to use fewer server resources than other solutions. Since licensing is by CPU, reducing the number of CPUs necessary to sustain the data center yielded a 65% reduction in server-related capex costs toward TCO.
Learning From the Best
Both Ericsson and IT Brand Pulse demonstrate sound methods for determining where TCO can be reduced. But building the most efficient data center will never be a simple case of changing one component. To see results like what these companies achieved, you have to understand the thinking behind their performance.
Each company in these examples invested the time to forecast the unique needs of its project. What are some examples for a fledgling data center?
- Physical space. Ericsson’s example shows how an innovative solution can place a lot of hardware in a confined space. Doing so can be cost effective, but it requires you to consider things like access for maintenance and the cooling needs of your hardware.
- Capex for systems. The IT Brand Pulse study investigated several commercial-grade storage solutions, but if you’re planning to build out a data center capable of hosting SaaS operations, you’ll need to own the hardware or strike up a massive contract with a major storage provider. It’s not just the cost of hard drive space that comes into play here. Customers won’t bite if your solution isn’t responsive and secure. Expect to invest in commercial-grade backup and cybersecurity solutions to instill confidence in your audience.
- Operational costs. A reliable system that costs 25% more but runs well and requires one person to manage is ultimately much cheaper than a system that comes at an affordable price but can’t run stably without constant attention from professional services or support. If you allow yourself to believe that buying the absolute cheapest solutions will lead to a lower TCO, expect to fail. The stress that unreliable systems place on a company incurs monetary and emotional costs that you’ll never make back, so pay a little extra for peace of mind.
Every solution is different, but some basic concepts apply to all tech enterprises. They apply to keeping the TCO low for your data center, too:
- Always think of the end user first. Choosing a system with fewer features that’s easier to use and cheaper to run might lead to a more profitable endeavor, even though you’ll know you could have done more. Features that seem easy to use to the experts on your team may not be viable to the general public, and they add cost and complexity.
- Break your deployment and ongoing enhancement efforts down into manageable projects. Forecast and track each one, recording the direct and indirect costs, capex and opex. Doing so will give you an accurate picture of spending over time, and it will allow you to see where you’ll need to account for extra costs or generate surplus funds.
- Get outside advice. It’s easy to become buried in your work and lose sight of things that are obvious for those who don’t live and breathe the stuff. Consultants and the general public can both offer valuable insight. Don’t undervalue the importance of a perspective from outside of your business.
- Overcommunicate to your staff. Keeping TCO down will mean making changes. People might not react to those changes in the way you want them to. That’s inevitable. But your people will react more favorably to changes they know are coming and understand are for a good cause. Let them know what to expect whenever you can, and they’ll give you valuable feedback and loyalty.
Living in fear of the bottom line is no way to run a business. The tools and the knowledge available to business owner/operators today grant you access to more of the critical data you need to be successful. Read the case studies and follow my advice—you can launch the business you’ve always dreamed of, or make your existing project more profitable.
About the Author
Kayla Matthews is a technology writer and reporter, contributing to websites such as VentureBeat, Vice, MakeUseOf and TechnoBuffalo. Visit ProductivityBytes.com to read more recent posts by Kayla.