In-house versus outsourcing IT operations and disaster recovery are two hot issues facing CIOs today. Both involve finding the proper balance among control, cost, security and reliability.
Unfortunately, IT organizations all too often become fixated on one of these factors and minimize the importance of analyzing all of these factors before making a decision.
There is no one best answer regarding how these factors are to be balanced so that corporate strategic and operating objectives are met. Technologies which allow for remote control of processing equipment and systems, combined with the weak economy, volatile capital markets, corporate mergers and current supply/demand imbalance in colocation capacity may well enhance the appeal of outsourcing primary, secondary and/or disaster recovery operations.
Once an organization decides to locate all or part of its IT operations offsite in a colocation processing center, choosing which colocation facility that best meets their requirements requires considerable analysis.
Separating The ‘Wheat From The Chaff’
Having managed several corporate IT operations that used offsite colocation, one thing is clear — not all colocation facilities are created equal.
The exponential growth of dot-coms and telecoms during the later half of the ’90s drove a building frenzy in data centers which outstripped the needs of these industry segments, as well as those of corporate America in general, particularly after the technology bubble burst. The result has been a glut of colocation, data center space with capacity utilization ranging between 30%-35% worldwide.
Taken at face value, the magnitude of data center overcapacity, given the slow pace of economic recovery and the disastrous state of the telecom and technology industries, would lead one to conclude it will be a buyers’ market for the next decade. In reality, however, data center quality and technical support vary widely, from the low-end converted warehouse with no raised floor and a fiber core to state-of-the-art, mission-critical facilities in which the user can move in with little if any capital investment and be assured of trouble-free operation.
The low-end space is an overbuilt non-differentiated commodity, which likely will be the last to be occupied or eventually transformed back to its original use. High-quality colocation space, on the other hand, is relatively limited, expensive to build ($500-$600 per square foot) and likely to reach supply and demand equilibrium in the next two to three years. I’ve found the web site www.carrierhotels.com to be an excellent source of information
for the industry.
Evaluating and Selecting Mission Critical Colocation Space
One of the first things to do when evaluating a data center is to understand what management’s objectives were in designing and operating a colocation facility. In particular, it’s my experience that what differentiates the top-quality data centers from the low-end commodity facilities is that they are designed, constructed and operated from day one with the objective of satisfying the current and future requirements of the most demanding clients.
Best-of-class colocation facilities provide flexibility in terms of scalability, connectivity and monitoring options; fail-safe security; a redundant, quality infrastructure with 99.999% reliability; and knowledgeable technical, monitoring, and support staff dedicated to meeting customer requirements from installation to ongoing operations. By meeting these requirements, a data center can offer users a cost-effective alternative to building and operating an equivalent quality facility in house.
In addition to understanding the design and operating philosophy, CIOs evaluating colocation options should ask about the firms that were employed to design and actually build the facility. The key here is to determine that the facility was designed by a top architectural firm and used the “best of the best” engineering, construction and systems firms. If that is the case, then it is likely that the data center will be capable of delivering mission-critical “five 9’s” reliability.
To help structure the process of evaluating a colocation data center, CIOs may find it helpful to rate a data center against critical criteria. A client considering relocating their disaster recovery and parts of their IT operations to Sacramento recently asked me to evaluate the new 92,000 square foot, fully built out, mission-critical Herakles Data Center.
As part of this process I spoke with the architectural company, which designed the facility, and the construction company that built it as well as the data center’s management. I then toured the facility and utilized a comprehensive site selection, building criteria, and best practices checklist to systematically evaluate the data center.
Herakles engaged Callison Architecture (one of the country’s top 10 architectural firms) and the engineering firm of Einhorn Yaffee and Prescott (EYP) (the designers of the NASDAQ building in New York) to design their facility. Joe Dillingham of EYP explained, “Our objective in designing the Herakles facility was to insure that the electrical and mechanical systems at Herakles were mission-critical in terms of reliability and redundancy. Equally important we wanted the systems to be able to be efficiently maintained so that there would be no fall-off in performance over time.”
The general contractor who managed construction of the Herakles data center was DPR Construction. DPR was chosen as the general contractor because of its experience and track record of building world-class data centers for such companies as E*TRADE, AT&T, and Charles Schwab. David Dovichi, DPR’s project manager, calls the Herakles Data Center “truly world class.”
“The 92,000 square foot raised-floor facility has N+1 redundancy for all electrical and mechanical systems; its mechanical and biometric security systems are cutting edge, and the building structure itself rock solid,” says Dovichi. “In fact, even though the data center is located in seismically neutral Sacramento, it is built to exceed state and regional seismic requirements. The 5,000 square foot Command Center looks like something out of a sci-fi movie and is capable of monitoring and controlling all aspects of the facility’s systems. Yet with all its technical sophistication the facility is aesthetically pleasing and practical providing users with office space, lab rooms for equipment staging and well-located receiving docks with storage cages to further assure security.”
Given the quality of vendors involved in designing and building the data center, combined with the fact that the goal from the start was to build a state-of-the-art, world-class facility, the stage was set for performing a detailed evaluation of whether Herakles in fact achieved its objectives.
The results of my evaluation are summarized in Tables A and B and are
consistent with the comprehensive site selection and building criteria
checklist from HLW (as reported by Gartner in TechRepublic), one of the foremost designers of large sophisticated
Table A — Physical Systems & Services*
Criteria category Number of criteria met Site/location 8 out of 8 Architectural 12 out of 12 Structural 3 out of 3 Mechanical engineering 2 out of 2 Electrical engineering 4 out of 4 Plumbing/fire protection 5 out of 5 Telecommunications 1 out of 1 Security Exceeds industry standards and end-user needs
* Consistent with criteria reported by Gartner in TechRepublic
Table B — Infrastructure & Backbone Support**
Criteria category Number of criteria met Facility and physical requirements 18 out of 20 Physical security 16 out of 17 Network security 13 out of 13 Operations 19 out of 19 Backbone connectivity 13 out of 14 Gateway/WAN edge layer 6 out of 7 Core layer 5 out of 7*** Distribution layer 4 out of 6*** Access layer 2 out of 2 Cabling 6 out of 6
** Criteria from Cisco Systems’ best practices checklist.
*** Two of the criteria are customer controlled.
Although “best practices” by definition is a term used as an ideal target and rarely met, I found Herakles to surpass the criteria listed in Tables A and B in a number of categories, such as raised floor, earthquake stability, building management systems, roof access security and monitoring, operations command center, business convenience for clients and partners, and shipping and receiving.
Bricks And Mortar Are Only Part of the Equation
While the Herakles facility met all the “bricks and mortar” criteria for being mission critical, basing a decision to relocate an IT operation primarily on “bricks and mortar” is fraught with risk and potential for unpleasant surprises.
Equally important to any evaluation is the financial and real estate arrangements which underpin a data center, the ongoing maintenance of the facility and most important the people who manage and operate the facility.
For example, many colocation facilities are not owned by the companies operating them instead they are leased from a third party for five to 10 years. As a result, users face the risk that the building lease may not be renewed or that future rents and costs will increase dramatically when the colocation operator renews the lease. Further, if an operator does not own the facility, they are likely to minimize investing for maintenance or facility upgrades.
Unfortunately, more often than not users are unaware of these risks when the operator does not own the colocation facility they use. In the case of the Herakles Data Center, the operator owns both the facility and the land, thereby eliminating the potential for real estate event risk.
Financial security of a data center is also important, especially since so many colocation data centers have recently gone bankrupt. While many of these bankruptcies have not resulted in users having to relocate their equipment, over time they may well impact the quality of service provided and/or facility maintenance. In evaluating the financial strength of Herakles, I was informed that the facility was profitable, cash-flow positive and the principles of Herakles had significant financial strength and access to both debt and equity markets.
Since uninterrupted service and 99.999% reliability is of paramount importance to the great majority of IT operations, how a facility is maintained is of vital importance. All too often management of new data centers take the position that since everything is new, facility and systems maintenance can be postponed for three to five years. This approach may save money in the short term, but it puts users at increasing risk of service interruption. I was pleased to learn that this was not the case with Herakles. Its management was taking a pro-active approach to maintenance. Utilizing Einhorn Yaffee and Prescott (the firm who designed and specified the electrical and mechanical systems) the company had developed and was implementing a comprehensive preventive maintenance program.
“We are investing over $250K in our first year of operations to insure against any future interruption in service resulting from a failure of our systems infrastructure,” said Lou Kirchner, Herakles’ CEO.
People Make The Difference
While designing and building a truly world class data center is a challenge, ultimate customer satisfaction depends not only on the system infrastructure but also on the quality, culture, and experience of the people managing and operating the facility.
All too often organizations doing due diligence on a colocation data center focus only on the building and its technology, and give short shrift to the human side of the business, especially in understanding the culture of a facility’s management in terms of customer service and quality.
Here, again, it is important to talk to all levels of management and technical staff and, if possible, other clients who are using the data center. In the case of Herakles, everyone I spoke with from the receptionist to CEO had one overriding focus: to meet customer requirements in a timely manner — the first time every time.
Selecting an off-site data center is always a high-anxiety undertaking given the importance of having uninterrupted access to one’s databases. This being the case, it is vitally important to conduct a multi-dimensioned analysis of a colocation data center, which goes beyond the bricks and mortar and infrastructure technology and systems.
Overlooking the people, culture, maintenance programs, whether a facility is owned or not and its financial stability can result in unpleasant surprises. Given the current environment and excess colocation capacity, it is tempting to overweight such factors as cost and buildings and underweight what really makes them deliver five 9’s mission critical services over the long term.
Barry N. Shufeld has more than 30 years’ experience in IT. Since 1989 he has served as SVP and CIO for Finlay Enterprises, Triarc Companies, and led security and information systems strategic planning for Rhone-Poulenc. During this period Shufeld headed the building, relocation, and management of multiple data center facilities. Prior to that, he led IT auditing for International Paper Co., which included reviews, and audits of data centers globally. Currently principal of BNS Associates, Shufeld received his B.S. in
Information Systems Management from the University of Maryland Robert H.
Smith School of Business, and MBA in Information Technology Management
from Fairleigh Dickinson University.