Industry Standard

Tier Classifications Define

Site Infrastructure Performance
By W. Pitt Turner IV, P.E. and Kenneth G. Brill

One of the most common sources of confusion in the field of uninterrupted uptime is what
constitutes a reliable data center. All too often, reliability is in the eye of the beholder. What is
acceptable to one person or company is inadequate to the next. As Internet website hosting
continues to grow into a major industry, competing companies with data centers of radically
different infrastructure capabilities are all claiming to deliver “high availability.”

With the explosive growth of the Internet comes an increased demand for computer hardware
reliability. Information technology customers expect reliability of “Five Nines,” or 99.999%.
Unfortunately, the substantial investment a business makes to achieve “Five Nines” in its
computer hardware is insufficient to protect its mission-critical computing functions. These
significant investments must be accompanied by a solid understanding of how well their site
infrastructure can support their availability goals.

The Uptime Institute® has developed a tiered classification approach to site infrastructure
functionality that addresses the need for a common benchmarking standard. The Institute’s
system has been under development for several years, and includes measured availability figures
ranging from 99.67% to more than 99.99% It is important to note that this range of availability is
substantially less than the current Information Technology (IT) expectations for “Five Nines.”

Over the last forty years, data center designs have Dual power technology requires having at least two
evolved through at least four distinct stages, which completely independent electrical systems. These
are captured in the Institute’s classification system. dual systems supply power via diverse power paths
Tier I first appeared in the early sixties, Tier II in the to the computer load, which moves the last point of
seventies, Tier III in the late eighties and early electrical redundancy from within the
nineties, and Tier IV in 1994 with the United Parcel Uninterruptible Power System (UPS) down to within
Service Windward project, which was the first site to the computer hardware itself. Brill’s intuitive
assume the availability of dual-powered computer conclusion has since been confirmed by Uptime
equipment. The Uptime Institute participated in the Institute research that has determined that 95% of
development of Tier III concepts and pioneered the all site infrastructure failures occur between the UPS
creation of Tier IV. and the computer load. Since completion of the
Windward project in 1994, System+SystemSM Tier IV
Invention of Tier IV was made possible by Ken Brill, electrical designs have become common and the
Executive Director of The Uptime Institute, who number of computer hardware products with dual
envisioned a future when all computer hardware inputs has grown.
would come with dual power inputs. During
construction of the $50 million Windward project, The advent of dual-powered computer hardware in
United Parcel Service worked with IBM and other tandem with Tier IV electrical infrastructure is an
computer hardware manufacturers to provide dual- example of site infrastructure design and computer
powered computer hardware. hardware design simultaneously achieving higher

1
 The Uptime Institute
Industry Standard
Tier Classifications Define
Site Infrastructure Performance

availability. With the significant improvements in Such sites are classified Tier IV electrically, but only
computer hardware design currently being made, achieve a Tier II level mechanically. The following list
many data centers constructed even in the last five summarizes the characteristics of each Tier.
years offer only Tier I, II, or III functionality, falling
far behind in their capacity to match the availability ! Tier I
offered by the information technology they support. Single path for power and cooling distribution, no
redundant components, 99.671% availability.
Defining the Tiers
The tier classification system involves several ! Tier II
definitions. A site that can sustain at least one Single path for power and cooling distribution,
“unplanned” worst-case site infrastructure failure redundant components, 99.741% availability.
with no critical load impact is considered fault
tolerant. A site that is able to perform planned site ! Tier III
infrastructure activity without shutting down critical Multiple power and cooling distribution paths, but
load is concurrently maintainable (fault tolerance only one path active, redundant components,
level may be reduced during concurrent concurrently maintainable, 99.982% availability.
maintenance). It is important to remember that a
typical data center site is composed of at least twenty ! Tier IV
major mechanical, electrical, fire protection, security Multiple active power and cooling distribution paths,
and other systems, each of which has additional redundant components, fault tolerant, 99.995%
subsystems and components. All of these must be availability.
concurrently maintainable and/or fault tolerant for
the entire site to be considered concurrently The availability numbers have been drawn from
maintainable and/or fault tolerant. industry benchmarking conducted by The Uptime
Institute and sites in the top 10 percent (this means
Some sites built with fault-tolerant System+System only 10% of all sites performed at this level). The
electrical concepts failed to incorporate the mechanical quality of human-factors management is the most
analogy, which involves dual mechanical systems. significant element separating top sites from all others.

2
 The Uptime Institute
Industry Standard
Tier Classifications Define
Site Infrastructure Performance

Tier I Data Center activities include preventive and programmable

Basic maintenance, repair and replacement of
A Tier I data center is susceptible to disruptions from components, addition or removal of capacity
both planned and unplanned activity. It has computer components, testing of components and systems,
power distribution and cooling, but it may or may and more. For large sites using chilled water, this
not have a raised floor, a UPS, or an engine generator. means two independent sets of pipes. Sufficient
If it does have UPS or generators, they are single- capacity and distribution must be available to
module systems and have many single points of simultaneously carry the load on one path while
failure. The infrastructure should be completely shut performing maintenance or testing on the other path.
down on an annual basis to perform preventive Unplanned activities such as errors in operation or
maintenance and repair work. Urgent situations may spontaneous failures of facility infrastructure
require more frequent shutdowns. Operation errors components will still cause a data center disruption.
or spontaneous failures of site infrastructure Tier III sites are often designed to be upgraded to
components will cause a data center disruption. Tier IV when the client’s business case justifies the
cost of additional protection.
Tier II Data Center
Redundant Components Tier IV Data Center
Tier II facilities with redundant components are Fault Tolerant
slightly less susceptible to disruptions from both Tier IV provides site infrastructure capacity and
planned and unplanned activity than a basic data capability to permit any planned activity without
center. They have a raised floor, UPS, and engine disruption to the critical load. Fault-tolerant functionality
generators, but their capacity design is “Need plus also provides the ability of the site infrastructure to
One” (N+1), which has a single-threaded sustain at least one worst-case unplanned failure or
distribution path throughout. Maintenance of the event with no critical load impact. This requires
critical power path and other parts of the site simultaneously active distribution paths, typically in a
infrastructure will require a processing shutdown. System+System configuration. Electrically, this means
two separate UPS systems in which each system has
Tier III Data Center N+1 redundancy. Because of fire and electrical safety
Concurrently Maintainable codes, there will still be downtime exposure due to fire
Tier III level capability allows for any planned site alarms or people initiating an Emergency Power Off
infrastructure activity without disrupting the (EPO). Tier IV requires all computer hardware to have
computer hardware operation in any way. Planned dual power inputs as defined by the Institute’s Fault-

3
 The Uptime Institute
Industry Standard
Tier Classifications Define
Site Infrastructure Performance

Tolerant Power Compliance Specification Version availability offered by any data center, and are
2.0 (www.uptimeinstitute.org/spec.html). particularly important in a “Four Nines” Tier IV data
center housing IT equipment requiring “Five Nines”
Tier IV site infrastructures are the most compatible availability.
with high availability IT concepts that employ CPU
clustering, RAID DASD, and redundant Authorship
communications to achieve reliability, availability, Pitt Turner is a professional engineer, a
and serviceability. The accompanying chart shows distinguished fellow of The Uptime Institute, and a
how these IT ideas relate to site infrastructure principal in Computersite Engineering®. He has
concepts. guided over $1.5 billion in site infrastructure
investment for primarily Fortune 50 clients. Ken Brill
Solving Incompatible “Five Nines” is Executive Director of The Uptime Institute, and a
Expectations principal in Computersite Engineering. He is the
Even a fault-tolerant and concurrently maintainable founder of the Site Uptime Network® and invented
Tier IV site will not satisfy an IT requirement of “Five dual power distribution technology in 1991 for high
Nines” (99.999%) uptime. The best a Tier IV site can availability data centers.
deliver over time is 99.995%, and this assumes a site
The Uptime Institute is a pioneer in creating and operating
outage occurs only as a result of a fire alarm or EPO,
knowledge communities for improving uptime
and that such an event occurs no more than once
effectiveness in data center Facilities and Information
every five years. Only the top 10 percent of Tier IV
Technology organizations. The fifty-one members of the
sites will achieve this level of performance. Unless
Institute's Site Uptime Network are committed to achieving
human activity issues are continually and rigorously
the highest levels of availability with many being Fortune
addressed, at least one additional failure is likely
100 companies. They interactively learn from each other as
over five years. While the site outage is assumed to
well as from Institute sponsored meetings, site tours,
be instantaneously restored (which requires 24 x
benchmarking, best practices, uptime effectiveness metrics,
“forever” staffing), it can still require up to four
and abnormal incident collection and trend analysis. From
hours for IT to recover information availability.
this interaction and from client consulting work, the Institute
prepares white papers documenting Best Practices for use
Tier IV’s 99.995% uptime is an average over five
by Network members and for the broader uninterruptible
years. An alternative calculation using the same
uptime industry. The Institute also conducts sponsored
underlying data is 100% uptime for four years and
research and offers insightful seminars and training in site
99.954% for the year in which the downtime event
infrastructure management.
occurs.
This white paper may be quoted, reproduced, or distributed
Higher levels of site uptime can be achieved by in its entirety at no charge. The Uptime Institute exclusively
protecting against accidental activation or the real reserves the right to certify and determine the tier ranking of
need for fire protection and EPOs. Preventatives site infrastructures as defined in this white paper. The
Institute's comprehensive site certification process involves
include high sensitivity smoke detection, limiting
significant additional criteria beyond the summary level in-
fire load, signage, extensive training, staff formation provided in this document. Sites reviewed and cer-
certification, limiting the number of “outsiders” in tified by the Institute can be seen at www.upsite.com. This
critical spaces, and treating people well to increase white paper is posted and maintained on The Uptime Institute’s
pride in their work. All of these measures, if taken, website at www.upsite.com/TUIpages/tuiwhite.html.
can reduce the risk of failures. Other solutions
include placing the redundant parts of the IT © 2001 The Uptime Institute
computing infrastructure in different site
infrastructure compartments so that a site
infrastructure event cannot simultaneously affect all
IT systems. Another alternative is focusing special
effort on business-critical and mission-critical
applications so they do not require four hours to 1347 Tano Ridge Rd, Santa Fe, NM 87506
restore. These operational issues can improve the Fax (505) 982-8484 Phone (505) 986-3900 E-mail [email protected]

4
0203 TUI 705