Data Center Configuration Best Practices You Should Know

Data centers play an increasingly important role in enterprise IT, with most computing action now taking place in the data center.

While analysts at Gartner predict that enterprises will close downOpens a new window 80 percent of their traditional data centers by 2025, businesses also are starting to move away from their public cloud strategy; an IDG reportOpens a new window commissioned by data center consultancy, Insight, found that 84 percent of businesses reported recently moving some of their workloads off of the public cloud.

So it is interesting times for the data center, with a hybrid strategy on the rise that includes public, private and even on-premise workloads.

For businesses with plans beyond just the public cloud, here are seven data center configuration best practices that IT departments should keep in mind.

1. Stress Performance, Not Cost

Keep first things first.

“Focus on power and network optimization, even at the expense of cost and space,” says David Linthicum, chief cloud strategy officer for Deloitte ConsultingOpens a new window .

“Is it important for businesses to make sure they are optimizing acquisition costs for every infrastructure element, or do they want to prioritize for a solution that will work?” adds Kaustubh Das, vice president of product management for networking giant, CiscoOpens a new window . “What about paying for automation that will free up bandwidth or engineering that will buy them optionality?”

2. Make Power and Network Optimization a Priority

Foremost among those performance considerations are power and network optimization.

“Power and network optimization are typically not considered,” notes Linthicum. “Indeed, businesses think that those are resources that are always going to be constant.”

That’s not the case, however. Data centers can burn up to 40 percent more power by putting in a small rack on a larger unerupted power supply, according Linthicum, and networks can be saturated easily if not architected properly. So pay attention to the little details related to power and networking, because these details can have outsized impact even if the most efficient solutions cost a little more.

3. Allow for Agility Through Plug & Play Configuration

Configure your data center with scalable, modular infrastructure because everything will change in a couple years and a lack of modularity limits the adjustments that can be made.

“A typical data center should be expected to have a life expectancy of at least 15 years,” says Phil Rafferty, president of data center design firm, Data Specialties IncOpens a new window . “With the constant changes and trends in information technology, it is difficult to forecast what infrastructure is required much beyond 3 to 4 years.”

So to prevent over-building the data center, use designs that are modular and that can scale up as the load increases.

“This way, a facility only has the expense of constructing what they need plus two to three years worth of growth,” he explains.

4. Seek Complete Redundancy

Data centers must have redundant components and multiple distribution paths serving all equipment. Redundancy is the lifeblood of a high-performing data center.

“Configure for redundancy so you can change out many downed servers at the same time, and there is no need to pull them out of the racks each time one fails,” advises Linthicum.

This extends to facility redundancy, too.

“We have seen countless data centers where poor planning led to the use of temporary air conditioning units, open doors, and other near-disasters just to keep systems from overheating,” notes Tom Johnston, principal architect for consulting services at cloud and data center consultancy, InsightOpens a new window . “Modern server architectures are physically dense and require much more cooling and power than servers from just 10 years ago.”

5. Lean on Hot Aisle/Cold Aisle Containment Systems

Speaking of heat management, arrange equipment configurations so containment systems can lower operating costs and allow data center operators to run their system closer to its recommended design capacity.

“The IT equipment racks must be arranged in hot/cold aisles and either hot or cold aisle containment systems utilized to prevent air mixing and improve the efficiency of the mechanical system,” stresses Rafferty. “Containment systems should include installing blanking panels inside the racks, providing cold lock devices in the raised floor, and sealing gaps between and at the bottom of the rack.”

6. Automate and Leverage Infrastructure as Code

Infrastructure as code (IaC), which automates configuration and provisioning through machine-readable definition files rather than physical hardware configuration or interactive configuration tools, both sets the right foundation for a data center and makes it more dynamic as needs change.

“Implementing automation with IaC helps to builds in best practices around data center operations from the start,” says Johnston. “Automation normalizes and standardizes your environment so that it takes less effort to manage and is more secure.”

Additionally, IaC helps promote other goals such as scaling, proper governance, service catalog delivery and self-service, according to Johnston. It helps promote business control of consumption through quotas and show backs, too.

7. Configure a Virtual Cockpit with Full Control

Finally, IT should configure a data centers with a cockpit of controls that go all the way from applications to hardware infrastructure. From this cockpit, it should be able to both monitor and adjust all aspects of data center operations.

“It builds an integrated view of the data center,” says Das. “This visibility is powerful, and allows for a common data set between teams, as well as a common vocabulary and integration into the native tooling for each of the platform, infrastructure and app teams.”

The data center is changing. With these best practice, however, data centers can stay nimble and adjust as technology and needs evolve.