Why data centre moves are an opportunity for service providers

Why data centre moves are an opportunity for service providers
With the most experienced analyst team in the industry, Ovum Telecoms is uniquely positioned to provide the insightful analysis needed to guide service providers, regulators and vendors through the ma

Ian Redpath, Principal Analyst, Network Infrastructure

[email protected]

Internet content providers, stock exchanges, banks, communications service providers (CSPs), and others are investing billions of dollars to build data centre space. The proliferation of data centres in new geographic locations is fundamentally changing network demands and driving larger volumes of traffic. Data centres (DCs) with thousands of servers will require terabits of optical transport. CSPs have an opportunity to serve these new high-capacity transport network demands.

Data centre headlines and new connectivity required

Data centre investment announcements have been impressive over the past few years. Recent expansion activities include the following:

  • Google invested €450m for a DC in Hamina, Finland, 150km from Helsinki
  • ATT built a $200m DC in Kings Mountain, North Carolina, 50km from Charlotte
  • IBM invested $1.2bn in 15 DCs, including one in Barrie, Ontario, 100km from Toronto
  • NYSE built a 400,000-square-foot DC in Mahwah, New Jersey, 50km from Manhattan
  • Banco Santander opened an 85,000-square-meter DC 100km from Sao Paulo
  • Digital Realty Trust expanded its optical network by over 100km to interconnect its suburban London DCs beyond the M25 in Woking, Redhill, and Crawley
  • NTT Communications built a 400,000-square-foot DC at Cyberjaya, 30km away from central Kuala Lumpur
  • the Singapore Government built a “data center park” 30km away from the city center.

Each new data centre requires connectivity from the new site back to the large cities and into the network core. New network demands with a significant distance component represent new opportunities for the service provider community.

Data centre site selection criteria

Why are data centre operators selecting suburban and remote sites for new builds? Site selection criteria include:

  • Power: access to low-cost, diversely supplied, reliable power, with a climate conducive to minimizing power costs
  • Fiber: latency, access to diverse fiber, international cable landing hubs
  • Site costs: real estate costs and government incentives
  • Site diversity: “out of flood plain,” “out of seismically active region,” far enough away for site diversity but close enough for access.

Each data centre operator’s site selection calculus depends on its individual business model. Managing power costs is the principal motivator for selecting a more remote site.

Segmenting data centre operators (DCOs)

The optical network (ON) community needs a deeper understanding of data centers as future ON growth becomes more intertwined with data centre growth. Tracking, classifying, and segmenting data center types and understanding their preferred method of interconnection, their underlying motivation, and their position in the value chain is critical. Ovum has segmented the DCO market into five categories:

  • CSPs such as AT&T, China Telecom, Colt, NTT Communications, and Verizon
  • Internet content providers (ICPs) such as Amazon, Apple, Facebook, and Google
  • Carrier-neutral data center providers such as Equinix, Interxion, and TeleHouse
  • Financial entities such as Bank of America, Banco Santander, and NYSE Euronext
  • Other, including non-financial enterprises and governments.

Each type of DCO has a different business model and motivation for owning and operating data centres, a different set of network assets, and a different approach to data centre interconnect (DCI). The CSP opportunity for data centre interconnect will be unique for each segment.

Data centres have evolved from the city centers to the suburbs

Data centres were initially deployed in city centres such as central London and London’s Docklands. As traffic increased, CSPs and DCOs began to deploy more equipment in the space-constrained facilities, and more companies sought access, adding further pressure. The immediate solution was to add new square footage outside city centres and rely on optical network connectivity to connect back to core sites. Over 300 data centers are now in operation in London and its suburban periphery.

The London case has been replicated in major data center cities worldwide. New York’s data centre market includes significant properties across northern New Jersey. The Washington, DC, market includes a multitude of suburbs and satellite towns in northern Virginia.

Optical network connectivity is required to interconnect the data centres across substantial metro areas. The metro optical network constantly evolves as new data centres are added, new carriers enter the market, new diverse backup paths are desired, latency-optimized routes are required, and new enterprises connect directly to the data centres.

Data centres are being deployed in remote, low-cost sites

For latency-tolerant data centres, immediate proximity to population centres is not required, but access to low-cost and reliable power is imperative. A data centre “cluster” has sprouted in central Washington State astride the Columbia River, the nation’s premier hydroelectric power source. Amazon, Google, Microsoft, Sabey, and Yahoo all draw upon Columbia River–supplied hydroelectric power.

The low-cost, remote-site model has been replicated to some degree in the US and Europe. Remote data centres are been commissioned in North Carolina, Iowa, Scandinavia, and Ireland, all leveraging low-cost real estate and power.

New connectivity requirements and optical network architectures

New network requirements are also emerging from the data centre sector. DCI is transitioning from 1–10Gbps to 10–100Gbps. Total system capacity is evolving from older-generation Gbps systems to today’s multi-Tbps systems.

Existing optical networks are built in metro, regional, and long-haul network layers. DCI can cross multiple network layer boundaries with costly hand-offs between layers. Networks designed and purpose-built for data center interconnect will be optimized to minimize intermediate termination and network equipment ports, minimize latency, and support rapid service turn-up.

Data centre demands may be episodic, event-driven, and unpredictable. Networks that are flexible and can support more dynamic traffic patterns will be in a position to out-compete less flexible, legacy network structures.

CSPs need to track the data centre market to stay ahead of the curve

The US leads in data centre market development, but other regions are following suit. The Western European market is also well developed but with some different characteristics. Data residency needs will ensure data centers exist in every country, leading to a proliferation of smaller-scale data centres. The Asia-Pacific region has a rapidly growing data centre market in its key world cities, and the overall market still has significant upside potential. Latin America’s data centre market is ramping up, led by regional financial institutions and incumbent CSPs.


The analysis in this report is derived from primary research in the form of discussions with service providers, data centre operators, and equipment vendors. Secondary research included tracking press releases, public announcements, conference presentations, and other publications.

The post Data center move is an opportunity for service providers appeared first on Ovum.

View Comments
Leave a comment

Leave a Reply

Your email address will not be published.