Are Centralized Datacenters Distressing? Make Way for Micro-Datacenters

Change is the life-force of every organization. The unabated collision of digital business initiatives with the legacy IT systems due to the pursual of digital transformation will never come to an end. Reliance on data center environments is continuously growing and organizations are thriving to transform their knowledge warehouses into a more flexible, simpler, and more efficient one. However, dealing with data centers sometimes bring new concerns to CIOs such as, should they focus on standardization and emulate with IT strategies or is it the time to experiment with flexibility? Yet, neither one would be a correct alternative as CIOs need to deliver both.

Wads of large enterprises haul their remote data center to one/few central position/positions. Moving away from data centers that contain numerous storage and network devices, and servers that require careful implementation, maintenance, and purchasing, enterprise leaders have realized the advantages of keeping a distributed group of data center—called “micro-datacenters.” As a plain concept, it is easy to deduce that a fleet of micro data centers could annex the latency and resiliency of one “centralized data center.” However, the argument on whether to continue with a centralized data center or stretch out by deploying micro data centers is not going to end anytime soon.

Stay ahead of the industry with exclusive feature stories on the top companies, expert insights and the latest news delivered straight to your inbox. Subscribe today.

Mapping and Humoring

Approximating data centers to consumer-end is given a term “edge computing”, as the micro-data centers are located close to end-user populations.

Do you remember AOL Inc. acquiring content-producing websites including The Huffington Post, Engadget, and the Patch? The firm has embraced edge computing to spin out server capacity quickly. It has successfully established rack-sized micro-datacenters so as to transform its IT infrastructure; the unmanned IT facilities are being managed remotely.

For traditionally large enterprises with colocation facilities, the deployment of equipment in those cages is somewhat a comparable approach. It will surely help them to chase the same decentralized, agile data center back end as AOL pursue.

The focus is to improve speed and increase reliability at the end-user face. Employees and consumers are often spread out to wide geographical locations, and it becomes a challenge for enterprises to drop width-intensive information—streaming content, complex architectural diagrams, medical images, etc—quickly to consumer-end. Enterprises can minimize this time lapse by offering micro-datacenters where information is stored closer to the user and delivery of data becomes faster. A simulation-rendering firm might set up data center inside each of the branches across the country instead of centralizing it at headquarters to host bandwidth-hogging data, for example.

How different is the Micro-datacenter Approach

As a business grow, the cost for its data center space and maintenance grows, which demands the deployment of increased number of servers. The economies of scale—in terms of operational and purchasing expenses—advance at one location. The auxiliary areas that drive a data center to work efficiently become dearer, while hardware costs fall.

Working across significantly large and more complex data centers may require high-priced troubleshooting tools with more sophisticated management. In amends, the requirement for more physical space requests to allot more IT resources at multiple locations. Now, this presents management with a new challenge to track entire infrastructure and interlink all.

Downtime or slow network response is another driver to opt a decentralized system. Many enterprises just cannot afford to be offline even for a minute. A centralized data center approach brings in the perspective of is a single point of failure. Nobody’s fault, it is anticipated even with rack redundancy architecture, redundant power supplies, and localized safeguards as well. If the supreme system goes down, the entire business shut off.

Distributed data center locations, on the contrary, have the tenacity to provide a more reliable environment for principal processing tasks from several sites locations. However, the role of virtualization in to-the-edge data contingency cannot be ignored; it is capable of transferring a virtualized process off a collapsing server to any of the series of servers situated remotely. Because the system load is smoothly doled out to several data centers, an occurrence of a glitch in one machine won’t hinder the workload. In such scenarios, faulty systems are immediately isolated from the mains and are fixed without facing any downtime.

A survey by International Data Corp. revealed that servers—especially at large centralized data centers—make the businesses pay in cold cash for large energy bills than in sticker price during its lifetime. Some easily exhausts local energy grids completely. Yet, distributed data centers consume energy from multiple locations and thus, strikes less on a grid.

Go Smaller or Pay the Price

Most companies cite to keep abreast with disruptive competitors but, fail to realize that it is a narrow approach to success, anyway. The primary concern of companies today is not how should they keep up with their competitors, but how closely can they meet the demands of their own customers.

Moving data closer to the user eliminates many problems and can massively improve the slow network response time. Prominent players such as IBM, Dell, and HP have developed many lightweight, compact, and more mobile data servers. Further, suppliers such as Schneider Electric, Elliptical Mobile Solutions, and Silicon Graphics Inc. (SGI), are battling royal with self-contained micro-data centers.

Last year, a report from MarketsandMarkets predicted that the IoT-driven growth on the network’s edge will prompt the breeding of micro data centers, fabricating the sector worth $6.3 billion by 2020.