If you’ve ever purchased an off-the-shelf laptop, you’ll likely be slammed with bloatware. It’s the glut of mostly useless software that comes pre-installed on every machine, hogging resources and slowing down your cool new purchase. Fortunately, most bloatware can be easily uninstalled. On a similar note, and more insidious to your company’s bottom line is the excessive hardware that often comes with buying traditionally manufactured server platforms. It’s a silent but constant resources hog, impossible to remove and constantly eating away at your operating budget.
Traditional computer server manufacturers often try minimizing their own operating cost by creating standard platforms. The lower the SKUs, the faster and cheaper the manufacturing process generally is, along with more simplified logistics of handling their inventory. So for people looking to buy one of these platforms, traditionally they must look at a tradeoff between performance, pricing, and long term operating expenditure of the hardware.
The problem with buying a standard server platform is that a small business selling a simple SaaS app can end up buying the same setup as a company that is using the server for high performance computing. The disparate server needs means that if the standard platform happens to fit exactly the level of processing power and storage that your company’s needs, than you are in luck. But for the majority of companies, you’ll probably end up with more hardware than you actually need, because these standard platforms are designed for a wide range of needs rather than your specific one. And when you find yourself nearing the end of the spectrum for your current setup, upgrading could mean a dramatic shift in cost as the next step up offers way more storage or processing power than you actually need.
Yet it’s a fallacy that more performance will equate to high CapEx and OpEx. After all, with a standard server platform you are likely paying for hardware you don’t need, not to mention the high energy cost and long term OpEx associated with powering all that excess hardware. Companies like Facebook and Apple have long figured out that custom hardware not only results in higher performance, but overall lower operating costs as well. The Open Compute Project (OCP) is one example of these large enterprise companies working to make hardware that is more efficient, flexible, scalable, and cost effective. However, OCP isn’t available to small businesses that average 100-1000 nodes, selling mainly to larger enterprises requiring upwards of 10,000 nodes, and requires a tremendous amount of lead time for product delivery.
In fact, few of the businesses developing custom hardware gear them toward the average size data center. However, custom hardware developers do exist, and they are worth seeking out. After all, bearing a larger than necessary energy footprint from all the excess hardware is not only costly, but results in a heavy environmental toll as well. According to the EPA, volume servers are responsible for 68 percent of all electricity consumed by IT equipment in data centers in 2006. A study by the U.S. government found that in 2014 US data centers consumed about 70 billion kilowatt-hours of electricity, making up two percent of the country’s total energy consumption. A custom, turnkey server solution built for your specific server needs will maximize performance while greatly reducing the excess energy expenditure, which means lower long term OpEx.
In addition, while some businesses can expect higher initial cost for these customized server solutions, that may not always be the case. For example, if you are on the lower end of the storage and processing needs, a custom solution may turn out to be more cost effective that buying a standard platform with much more upgraded hardware than you actually need. The specialization from the software perspective, not to mention the ability to customize software on servers delivered by turn-key solution providers, offer businesses much greater flexibility, higher performance, and overall lower expenditure over the long run.
Niche manufacturers exist to fill a much needed gap between the OCP providers that service larger enterprises and the inefficient standard platforms that many smaller businesses are stuck with. Taking similar machines and similar high-quality components as traditional server manufacturers, they can design unique, customized server hardware that greatly reduces inefficiency, creating a solution that is better for business and better for the environment as well.
About the Author
Alex Lesser is Executive Vice President at PSSC Labs, makers of handcrafted big data and High Performance Computing solutions.