Over the past few months there has been a great deal of discussion in the IT community and media regarding data centers and the good and bad impact they have on businesses and consumers. There are a lot of positives around large-scale data centers – including CAPEX reduction, easier to scale as a company grows and peace of mind knowing a company’s data is protected and backed up.  However, there is another side of the coin.

It’s no secret that data centers are not very green and are quite expensive to operate – especially the larger centers – from a capital expenditure perspective.  As more data piles up, more hardware is deployed, growing the footprint and as a result, adding more power and cooling costs. This doesn’t take into consideration the operational challenges this poses. More data plus more systems plus less consolidation equals big headaches.

The world’s largest data center, the SwitchNAP, owned by Switch Communications, is a staggering 2.2m sq ft, which is about twice the size of the second largest data center. A data center that size uses more than 202,000 tons of cooling to keep the servers from overheating.

On top of the cooling costs, the data center uses significantly more energy just to keep disks spinning 24/7/365. But what can be done to minimize the CAPEX and OPEX associated with data centers?

First, we need to look at the three components that make up data centers:

Computing – Power and cooling costs associated with computing depends on what servers are used as each type has different power and cooling requirements. The advent of server virtualization is a great concept from a pure CAPEX perspective, however, running virtual machines and consolidating servers drives greater resource needs. Ultimately, there is very little that can be done here to reduce overall energy consumption and costs.

Networking – Networking equipment is archaic and hasn’t changed in years. This equipment is built to carrier-class specs and as a result consumes large amounts of energy. Because there is very little development in this area, there is very little that can be done to reduce energy costs in this component.

Storage – Most data centers are beholden to hard disk drives (HDD) for their storage, but because of the mechanical components associated with HDD storage, these disk drives drink up the most power. However, there is an alternative, which significantly reduces energy costs associated with storage while increasing performance – flash storage.

Out of the three components of the data center, storage is the only area where an alternative to current technology can be applied in order to achieve a meaningful improvement. Since flash storage has no moving parts, it not only requires significantly less to cool the unit, it increases performance at the same time – flash doesn’t have to wait for slow movement of mechanical parts in order to do its job. As a result of the increased performance, organizations are able to consolidate their multiple HDD servers onto a smaller number of flash storage arrays, which helps to reduce overall footprint and power costs in the data center.

For example, the City of Delaware, Ohio, reduced its application server footprint from 12 existing physical servers to four physical servers, eliminated the need for an intended but prohibitively expensive purchase of 50 additional servers, reduced power and cooling costs by 90% and reduced the total cost of ownership for data center rack space from US$60,000 annually to $600 annually – all by switching from HDD storage to flash storage in the data center. Not to mention the increase in overall performance - reduced access time to a permitting software application from 20-100 milliseconds to one-two milliseconds and reduced MS Exchange access time from 10 minutes to 30 seconds for a city attorney who had 10K e-mails in his inbox.

To move to this model, however, you need to change your business processes as it relates to expenditures in your data center.

The views expressed are those of the author.