Mannix O’Connor serves as director of technical marketing for MRV Communications’ Optical Communications Systems Division, and is an established member of the telecommunications community, serving as a founding member of the IEEE 802.17 Working Group and co-chair of the Access Group for the MEF.
There is no commonly accepted definition of a “data center.” If you ask five different data center managers how to define the term, you’ll get five different answers. What we can all agree on is that, no matter how you view the concept of a data center, these centralized repositories for the storage, management and dissemination of information are characterized as having significant computing power, massive amounts of digital storage and sophisticated local area data switching networks.
However, all data centers share several challenges as well. In order to keep up with the constantly evolving mobility trends, next-generation applications and services, as well as common real estate, security, cooling and power constraints, new data center business models are emerging. We identify three categories of data center business models: compute-focused, full service and disaggregated. Each model refines its connectivity methods and operations to best serve its targeted customers. We’ll review each model, it characteristics, benefits and constraints in order to determine which might best fit your own power, storage, space and bandwidth needs.
Data Center Model 1: Compute-Focused
In the compute-focused model, the data center delivers IT resources and on-demand services such as servers, storage, operating systems and application processing through a virtual computing environment.
Compute-focused data centers have sophisticated networks within their premises for rapid machine-to-machine communications, but do not own or control an external wide area network (WAN). Data center personnel rely on other service providers to deliver and manage network connectivity to customers. They strive to have as many connections as possible to other providers in order to solicit business from the widest variety of customers.
One of the benefits of this model is that IT resources can be rapidly expanded and contracted to meet individual corporate objectives. This model is driving the rapid expansion of public data centers open to one and all, as well as private data centers that are part of a single organization.
Data Center Model 2: Full Service
The full-service data center model is represented by a WAN operator with one or more data centers within its network. In this model, the data center satisfies customers’ requirements with their own network and resources. Organizations with this type of structure range from dark fiber builders, to competitive local exchange carriers (CLEC) and incumbent providers.
In the full service data center, access is their trump card. Full service data centers often have a large access network to leverage and may be less eager than the compute-focused data center to encourage a wide variety of interconnections with other providers. The combined offering of the WAN and access network are often the strategic benefit these data centers deliver to their customers.
Data Center Model 3: Disaggregated
Historically, data centers consisted of large facilities with thousands of machines and support for millions of customers. The disaggregated data center model takes a more innovative and efficient approach to data center architecture.
In this model, variables such as content, raw compute power, storage and the network interconnection locations can be distributed geographically to optimize the cost and performance of applications. For example, content may be cached closest to the customers in large urban areas in order to improve response time, but this has the drawback of higher space and power costs in metropolitan areas. Hence, heavy computing power and tier 3 storage is increasingly being built in locations that have access to inexpensive or clean power and cheap real estate—often where a high percentage of electric capacity is generated from hydroelectric sources.
In addition, network access can be disaggregated as well. Connectivity costs for network access can vary widely depending on location and, consequently, disaggregated data centers may distribute network connectivity across a number of interconnection locations. In this architecture the remote locations with processing and storage are “tethered” to the central connectivity points via dark fiber.
It’s safe to say the likelihood of “data center” ever finding an industry-wide definition is slim to none. The previously mentioned business models represent how today’s data center has reformed itself to meet subscriber demand. However, the continuous arbitrage and cost differences between labor, power, network resources and real estate will continue to push the evolution of data center business. Luckily, we now have a variety of options to choose from, and can remain confident we are deploying or subscribing to the most effective model for our business needs.
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.