Brian Lavallée is the Director of Product & Technology Solutions at Ciena.
For a long time, businesses located their data center facilities in urban metro locations to be as close to their business points as possible to improve access performance to end users. But this is changing.
In fact, network operators and content providers to colocation operators, hosting and cloud-based service providers have started to build data centers further away from their customers due to improved costs related to real estate, power, and cooling, in order to lower operational costs and enhance geographic resiliency. For instance, Google, Yahoo, Dell, Microsoft, and Amazon have all built large-content data centers in the rural town of Quincy, Washington, which has access to several nearby hydroelectric dams that deliver low-cost green power.
This rise in remote data centers has helped enterprises cut costs and trim budgets significantly, but not without its share of networking challenges.
Let’s start by examining the triggers for the exodus of data centers from urban metro areas to remote locations. Data centers spend a considerable amount of money on hardware – upwards of a billion dollars in some cases – with additional associated taxes and fees. One way to keep costs in check is to minimize real estate expenditures. Savvy state legislatures, such as the one in Oregon, are aware of this and working to provide tax incentives for enterprises to relocate data centers to rural communities with available land, drier climate, and access to lower-cost power.
Another major operating cost relates to energy. Data centers are one of the largest and fastest growing consumers of electricity in the United States. According to a report from the Natural Resources Defense Council (NRDC), in 2013, U.S. data centers consumed an estimated 91 billion kilowatt-hours of electricity — enough to power all the households in New York City twice over. To minimize power usage, rural locations like Quincy have become much more attractive due to a climate that supports less costly renewable power sources such as hydro-electric power.
Thirdly, geographic diversity is fostering the movement to rural-based data centers. In the case of a power outage, organizations maintaining secondary data centers on different power grids from the primary location create added protection in avoiding a complete blackout. Also, areas prone to natural disasters, including earthquakes, floods, hurricanes, and tornados, are more reliably served by building data centers in different geographic zones to reduce catastrophic risks – natural or manmade.
Changing Dynamic for Data Center Interconnect (DCI)
The advantages of moving data center facilities to locations that provide lower-cost space, cheap energy, and geographic diversity are quite evident. However, as data centers are located further away from the urban centers that they serve, legacy optical network designs that demarcate metro and regional networks hinder performance and lead to cost-ineffective connectivity. To make remote data centers sustainable over the long term, it is imperative that traditional metro-regional domains evolve into seamless “user-to-content” and “content-to-content” networks. This allows the programmable network to dynamically connect end users – man or machine – to the required content and resources.
One of the challenges that must be addressed is increased intra- and inter-metro bandwidth requirements. The rapid growth of video-centric content is driving bandwidth requirements from 1 Gigabit Ethernet (GbE) to 10GbE and 100GbE services, which inevitably travels outside the metro network and onto regional and long-haul networks. In addition, enterprises are increasingly adopting a cloud computing and storage utility model, driving the need for on-demand connectivity that aligns with the emerging on-demand cloud consumption model. It’s very clear as data travels longer distances to end users that highly reliable and resilient network architectures are critical.
Another challenge brought on by crossing network boundaries involves inefficiencies. With the relocation of data centers into a regional network footprint connected via a backbone network, connectivity now has to span across an originating metro, through a regional network, and back through a terminating metro to reach the secondary data center. This back-to-back, cross-domain approach results in network inefficiencies due to multiple network demarcation handoffs, increased provisioning and engineering efforts for new services (including limitations in available service offerings), and a complicated process to manage end-to-end service performance.
By blurring the boundaries between metro-regional networks using seamless user-to-content and content-to-content network architectures, service providers can address the surging traffic growth across geographical metro-regional boundaries at distances up to hundreds of kilometers, or even further.
The Evolution to a Content-Based Environment
As traffic demands center more and more around user-to-content and content-to-content patterns, the traditional demarcation of metro and regional networks is now being reconsidered. It's especially the case with the advent of coherent optics and ROADMs that allow network operators to erase these historical, and now obsolete, demarcation points. It is now possible to create far simpler networks by removing the traditional seams resulting in less networking equipment, thus, lowering capital expenditures and associated power, space, and energy costs.
These optical networking advances ensure that as data centers move away from the urban city centers that they primarily serve, connections will be reliable, high-performing, and most importantly, cost-effective.
Industry Perspectives is a content channel at Data Center Knowledge highlighting thought leadership in the data center arena. See our guidelines and submission process for information on participating. View previously published Industry Perspectives in our Knowledge Library.