Data centers aren’t just in the basement anymore. In fact, they are not only linked to the traditional operational aspects of a hospital—personnel records and insurance information—they are now the critical link to the clinical side, as well. Newer technologies are allowing for remote access, x-rays are accessible from the doctor’s office, and worldwide video teleconference is becoming more commonplace. As a result, the data center is demanding more respect and attention.

For those hospitals and medical facilities about to upgrade or expand their data center capabilities, a more global approach may be required. How can a data center be master planned to work today and in the future when needs expand? Where is the best site? What level of reliability is actually needed? This article will examine key components of decisions that must be made when planning for a new data center.

The critical components

Today, hospital networks support all kinds of functions that may range from traditional accounting to highly advanced remote surgery. More and more lives depend on the information provided by the network. The increased use of digital medical imagery continues to increase the need for information storage that must be accessible in real time. These requirements force most hospitals to continually expand their data centers. Unfortunately, most hospital and institutional data centers aren’t able to expand because of poor planning.

Often, existing data centers are located in the basement of a hospital wing next to other building programs that are difficult to relocate. Many are poorly located for servicing and have very little security beyond a swipe-card lock. Often there are risky elements adjacent to these existing data centers, particularly drain pipes. In fact, one facility our firm recently worked with had acid waste lines from its labs located in the ceiling of the data center!

The typical existing computer room was designed around low-density computer and storage equipment. Today, space is at a premium and newer blade technology can put more computer power in a smaller box. While that sounds like the perfect answer, unfortunately there is a price that must be paid through the generation of more heat, which requires more cooling solutions—and the use of more power. This leads to hot spots and to power reliability problems if solutions are not put in place.

Another issue that plagues hospitals and institutions is the frequent location of departmental data centers all over campus (see sidebar, “Is There a Rogue Data Center on Your Campus?”). Often these legacy data centers and computer rooms occupy very valuable real estate in the center of campus, which can limit flexibility.

Once the institution realizes these issues are holding back progress and putting their current equipment at risk, it realizes a solution must be found. A professional consultant, engineer, or architect who focuses on data center design can help find this solution. These design professionals can create a working concept, schedule, and budget that then can be presented to hospital leadership so that funding can be obtained for a new data center project. (For information on the availability of these professionals, check with the Uptime Institute [] or attend a national data center conference, such as those sponsored by 24×7 Exchange or Afcom.)

Data center master planning

First, the team will come up with a concept of operation. An evaluation can be done on the current equipment and space to determine a growth profile projecting the hospital’s needs into the future in terms of servers, networking, and storage. This information can then be translated into a growth profile in terms of cooling load and power requirements. In addition, this information will be critical in evaluating when a new data center is needed and how much funding is necessary.

Another evaluation is needed to determine the level of reliability required by each system. The goal for a data center is to maximize reliability and minimize dollars spent, all while maintaining future flexibility. To do this most efficiently, each system should be supported to its level of recommended reliability—but no more.

Quite often reliability is expressed as “N” in Uptime Institute tier ratings. In simple terms, “N” equals need. So if a system is 2N, then the system has two completely separate systems that can support the network’s needs, i.e., the original, and a duplicate that serves as a backup system. The Uptime Institute translates these concepts into a tier rating system. These classifications are good rough guides, but quite often they are not a simple N+1 or Tier II solution.

There are several ways to achieve redundancy. The traditional way is to have one critical server supported by a very redundant mechanical/electrical infrastructure with some type of inactive backup. For hospitals and others, a dual-active approach seems to be more effective (figure). Under this approach, there are two mirrored, active servers separated from each other and supported by a slightly less redundant infrastructure. This approach may provide more reliability to the end user at a smaller price tag, since the price on redundancy grows exponentially as the tier ratings rise. Using this concept, hospitals can also get some value out of old data centers, which can act as a second active backup site to support the most critical applications. This will allow a facility to operate if there is a major problem with the communications grid.

For hospitals, a dual-active approach to redundancy seems to be more effective

These decisions will drive the size of the new data center. The growth profile will size the data center floor and the mechanical/electrical spaces. Other common data center spaces that should be considered are:

  • Receiving space

  • Burn-in space (used for unpacking equipment, loading and testing software, maintenance, decommissioning)

  • Storage

  • Tape storage

  • Staff desks and management offices

  • Help desk space

  • Vendor temporary desks

  • Telecom entry rooms and network rooms

  • 24/7 command center

  • Conference rooms

  • Facility shops and offices

  • Security rooms (for security staff to conduct central monitoring, monitor loading docks, locate security equipment racks)

If the data center is remote (off campus), the following spaces may also be needed:

  • Break rooms

  • Fitness center and other amenities

  • Lockers

  • Loading docks

  • Lobby

At this point, other goals can be established for the new data center, including using green building techniques or even achieving LEED ratings, which are becoming popular goals in new construction. A green data center may seem like an oxymoron, but LEED points in a data center can help greatly reduce energy costs.

Site selection

The site chosen for a data center can greatly affect its value to a hospital. Because on-campus real estate is probably tight, a remote data center is highly recommended. Today, connectivity is strong enough to “remote locate” even the most intensive applications. A perimeter site on campus, a site on an outpatient campus, or a totally separate site might be considered, as well. Benefits of a stand-alone, remote data center include:

  • Better security and standoff (i.e., distance from building façade to perimeter fence or other possible threat location)

  • Less risk of penetration (site can be unmarked)

  • Cheaper land

  • Preservation of valuable campus real estate for other purposes

  • No limit on other campus expansion plans

  • Better and cleaner power

  • Larger sites, which will allow for future expansion

When a remote data center is considered, real estate options open up. There are facilities that can host a data center (e.g., Sungard, Verizon Business, Rackspace) and developers (such as Digital Realty Trust) that will build a data center to suit and lease it back to the hospital. These options can be expensive, but the benefits may be worth the cost.

There are several key issues to evaluate for a new data center, whether it is remote or on campus:

  • Is there enough power? Proximity to quality, well-maintained electric substations is the most important issue. Power can be affected in various ways by neighbors, e.g., a data center should not be located near heavy industrial sites, since their machinery tends to create a dirty power situation that can shorten the life of equipment.

  • Can the information get there? Is there available dark fiber in the area? Can there be redundant, separate paths?

  • Is the site big enough? Can it accommodate the hospital now and in the future? Will it allow the creation of a good perimeter with some standoff?

  • Is water available? Cooling towers require huge quantities of water, so a reliable source is important. Municipal water is required, with some amount of water storage on site.

  • What are the risks to the site? Proximity to freight train lines, flood plains, and fuel refineries can expose the site to harm.

  • Can the staff get there? It’s important to choose a location convenient to the main campus where service to the facility will be easy. Proximity to current staff is imperative.

  • Is the site less expensive? Land in the suburbs may be cheaper then near a downtown campus. But are the power rates cheaper?

At the end of the master planning process, a hospital should have a good conceptual plan, site plan schedule, and budget. This plan should allow for large-scale growth while meeting the stated goals for the facility.


Data centers are indeed becoming the critical force behind how effectively and efficiently a hospital operates. As institutions expand, those that prepare carefully for the future will be positioned to succeed not only financially, but clinically, as well. This can only be accomplished by understanding specific needs, which comes from working with a team of experts and assembling the right team—including the architectural firm, the engineering firm, and the information technology specialists, among others. Those hospitals and medical facilities that do take this approach will be designing a successful path for information technology that will last for years to come. HD

Shawn Reichart is a principal in RTKL’s Baltimore office. As a project architect/project manager, he has worked at all levels of the design process, from concepts to construction administration. Since joining the firm in 1997, Reichart has contributed to RTKL’s portfolio of mission critical projects, most recently serving as project architect for the Highmark Data Center in Harrisburg, Pennsylvania.

Reichart can be reached at 410. 537.6000 or