ArticlePDF Available

Data Center Architecture Overview

  • National Academy for Planning and Development


Now is the era of cloud computing where internet based data are handled from remote places. Data is being entered. stored. processed, deposited and backed up all at the central servers located in specific buildings. Data center is a place where all these servers are gathered in compliance with art of technology. We have some limitations regarding knowledge about the architectures of data center. This article reveals some ideas about data center architectures and its components. Specially. this paper gives emphasis on explaining data center tier levels and its applications.
VOLUME 28, 2019
National Academy for Planning and Development
Development Revierv
Volume 28. 2019, ISSN: 1607-8373
Data Center Architecture Overview
Md. Abdur Rashid'
Now is the era of cloud computing where internet based data are handled lrom remote places. Data is being
entered. stored. processed, deposited and backed up all at tl.rc central servers locatcd in specilic buildings. Data
center is a place where all these servers are gathered in compliance with art of technology. We have some
limitations regarcling knolvledge about the architectures oldata center. This article reveals some ideas about data
center architectures and its components. Specially. this paper grves emphasis on explaining data center tier levels
ancl its applications.
Keywords: Data cenler. Cloud computing.
Information is power. Everyday information of any organization is increasing day by day.
World is becoming paperless. So cloud based data processing is going to be more popular.
But there are some questions on how the data centers. that we are imagining will be shaped.
What elements and standards of data centers will have to be maintained to attain future
data demand? It's worth mulling over on how we will be able to cope up with ongoing
technological changes and increasing scalability. Data center classification and its
standards, choosing tier levels u,iil help to make sustainable data center. This article will
help us to iearn about the definition of data center. its classification, standards, modularity.
fire system. network design. security, energy used and environmental (lTWatchDogs 2011)
issues.This article discusses some of the basic but key considerations lor planning and
designing data canters, which are dependent on the organization's business needs, tolerance
for failures trnd availability of budget. Lessons learned from this articie wiil be heipful for
researcher, academician, student and organizations dedicated to data center improvement.
Data center can be defined as (Rihards Balodis and Inara Opmane ) ' A data center is a
physical environment facility intended for housing computer systems and associated
components. Data centers comprise the above-mentioned computer systems and staff that
maintains them. The necessary physical environment facility encompasses power supplies
with the possibility to ensure backup power. necessary communication equipment and
redundant communication cabling systems, air conditioning, fire suppression and physical
security devices for staff entrances.'
Another way data center can be (Papaioannou20lT) defined as'Data center infrastructure
comprises a large number of servers interconnected by a network.'
In 1940, once thought that data center means, it holds a lot of computer inside it
(*center) is means a huge big computer rooms. ENIAC.
was one of the earliest examples data center. On that time, rt was a complex to operate the
computers and were needed many cables and other supporting devices with expensive
accessories. It requires big floor space to accommodate and needed racks to mount devices.
A single mainframe required a great deal of power and cooling was a big issue. During the
1 PhD Fello** oJ Bangladesh University of Professionals & Director (Adntin & Financel, Natiottol Acadetnt.for Planning
& De,-elopment (NAPD), Ministry of Planning, Bangladesh.
pAu l:
)< \ \ \\./ .&(7 \,/.-
Development Revielv
Volume 28. 2019
boom of microcomputer in 1980s, users tried to set at anywhere wjth little space and no care
about operating environment. With the advancement of Information Technology (IT)
operations more! user became more concern to maintain IT resources. UNIX operating
system became in 1970 after that free Linux operating system came in 1990. Bir using these
types UNIX and Linux computable PC operating system bought the term "Server". The
timesharing operating system rely on the client server model to laciiitate sharing unique
resources among multiple users. Inexpensive networking items couple with new standards
to put lT equipment's in a hierarchical design that prit the servers iu a specific room inside
the company. The use of the term "data center", as applied to speciaily designed computer
rooms, started to gain popular recognition about this time. It is noted in a article titled
History olData Centre Deveiopmelrt (Rihards Balodis) that the term data center itself has
been used since the 1990s.
List of commonly identilied benefits that can get from Data Center (DC) are:
. DC Reduce our information technology cost and risks.
. IT manpower requirement should be reduced together with the difficulty in retaining
staflwith IT ski1ls.
. DC provides high class data lacilit-v and state of the art network infiastructure.
. Immediate increase in the network reliability and scalability. As business grows and
requires more network resources; DC can easily meet new requirement.
. DC Managed Services provides you with readily available competent IT staffs for
operating, monitoring and maintaining system.
. DC Managed Services gives continuous operation. continuous support, security,
flexibility 24x7 days.
Data centers can be classified in different aspects depend upon its own characteristics.
Depending on the destination there are four diff-ei at types ( ) of data
1) Corporate data centers
2) Web hosting data centers, providing computer infrastructure as a service (IaaS)
3) Data centers that provide Turnkey Solutions
4) Data centers that use the technology to Web 2.0.
Data center energy expert Jonathan Koomey identified four types of data centers
(https ://
1) Public cloud providers (Amazon, Google)
2) Scientific computing centers (national laboratories)
3) Co-location centers (private 'clouds' where servers are housed together)
4) 'in-house' data centers (facilities owned and operated by company using the servers)
To evaluate the quality and reliability. data center standard (https://www.colocationam are measured. This standard also indicates the hosting ability of a data center.
The Uptime Institute ( is the IT industry's most trusted and
adopted global standard for the proper design, build and operation of data centers the
backbone of the digital economy. Uptime Institute tier standards provide a data center
classtfication also shown in following figure.
Data centers are classified in according to uptime are as follows :
. Tier 1 data center (99.611% minimum uptime)
. Tier 2 data center (99.'741% minimum uptime)
. Tier 3 data center (99.982% minimum uptime)
. Tier 4 data center (99.995% minimum uptime)
The higher the tier. the greater the availability. The data center tiers level, levels,
requirement and utilization places are (, www.netmagicsolutio shown in following table l:
Level Requirements Applications
1aSingle non-redundant distribution path serving the lT
N orrredundant capacity components
Basic site inlrastructure with expected availability of
Annual downtime 28.8 hours due to site
. Utilized by small
2Meets or exceeds all Tier 1 requirements
Redundant site infrastructure capacity components
with expected availability of 99 .l4loA
99.749% Uptime
Partial redundancy in power and cooling
Annual downtime 22 hours due to site
a. Utilized by medium
JMeets or exceeds ali Tier 2 requirements
Multipie independent distribution paths serving the
IT equipment
All IT equipment must be dual-powered and fully
compatible with the topoiogy of a site's architecture
Concurrently maintainable site infrastructure with
expected availability of 99.982a/o
No more than 1.6 hours of downtime per year
N+1 fault tolerant providing at least 72how
power outage protection
Annual downtime 1.6 hours due to site
Utilized by larger
4Meets or exceeds all Tier 3 requirements
A11 cooling equipment is independently dual-powered,
including chillers and heating, ventilating
and air-conditioning (HVAC) systems
Fauhtolerant site infrastructure with electrical power
storage and distribution lacilities wirh expected'
availability of 99.99 5%,
2N+ I fully redundant infrastructure (the main
difference between tier 3 and tier 4 data centers)
96 hour power outage protection
Annual downtime 04 hours due to site
Utilized by
Table 1: Data center Tier level
\ \ \ \./ ,w,
Development Revielv
Volurne 2[J. 2019
It is r-roted lrom above table that Tier 4 design requires double the inlrastructure of a Tier
IIi design. Also noted that both Tier III and Tier 4 data eenter specifications require IT
ecluipment to have dual power inputs. Following llgure 1 shows a typical tier 4 data ceflter.
Figure 1: Tier 4 data center
a. Technology infrastructure design
Data center design is the process of modeling an,.l designing (Jochim 2017) a data center's
IT resources, architectural layout and entire ilfrastructure. It enables the logical
conception of a data center prior to development or implementation in an organization or
IT environment.
b. Site selection
Forecasting next few years, data center location will be chosen. Some commonly identified
aspects (Yee Keen Seng) are proximity to available power grids, telecommunications
infrastructure, networking services, transportation lines and emergency services. These
aspects will help to operate data center smoothly.
c. Modularity andflexibility
Modularity (Ajay Ahuja 2015) and flexibility (Ajay Ahuja 2015) are key elements in
allowing for a data center to grow and change over time. Data center modules are
pre-engineered, standardized building blocks that can be easily configured and moved as
t "., -
iJlilifit' Pslv8r
t5l{lLJ A
: ; / I 1.. l','l. . i:,
2N Fully Redundonl
A!{ Fcth C*mP*rreni*
eep{lblc *f *uFPortirtg
I6O% si LoEd
t".. r:?1'{ti: t
6&ls &
\ \ ,/ \& Y,/././ x.
Data Center Architecture OVerview
d. Environmental control
The physical environment of a data center is rigorously controlled. Air conditioning is used
to control the temperature and humidity in the data center. Environmental factors
(ITWatchDogs 2011) like heat, hurnidity. airflow, smoke, and electricity can be equally
devastating to server room equipment's.
e. Electrical power
If national grid fails then a bank of batteries in a large data center, used to provide power
until diesel generators can stzrrt. Double feeder line can be established lrom near substation
to ensure uninterrupted power supply (Jochim 2017). Typical electrical power supply
system is shown following figure.
Utility Poser *1 Utitity Powe. f?
A ^ie.
l+\ b,
f. Fire protection
The primary objective is to protect human capital, data center resources and uninterrupted
IT services lrom any instant fire hazard. Detection, alarming, evacuation and extinguishing
(Siemens 2015) are the major concern in data center f,rre protects aspects. Two types of
measures are next page:
Fig 2: Cabinet aisle in a data center
.x. \\\\a ,,e/ \ ,/r
rj)iitii ilili ri:
{ -A'1
::rl.:r. li:r:.
ti ,iili::: li
.. t.. :.'.
/N//)s\ /.2^.
Development Review
Volumc lx. 2{)lq
/ 'Active measure
o Organizational measures and staff-related measures
o Structural and mechanical measures
o Technical measures
o Detection measures
o Alarm and evacuation measures
o Interventionmeasures
/ Passive (structural)measure
o Fire compartmentalizalron of a building.
o IJse of materials to prevent building collapse.
o Fire-resistant construction elements to limit spread.
o Provision offire-resistant escape routes/exits/staircases/elevators.
o Selection of materials to reduce fire load.
o Selection of materials to prevent toxic vapors.
Besides there is fire codes and standards (Hirnanshu 2014) lhal can be:
" 'NFPA 75: Standard for the Protection of Information Technology Equipment'
g. Security
Physical security also plays a large role with data centers. Physical access to the site is
usually restricted to selected personnel. Electronic access control and temporary visiting
card can be the solution tools to enter data center. Body checkup must be done before
entering data center. 24x7 days security must be ensured by professional security company.
Al1 security activities must be supervised by central administration.
h. Energy fficiency
There is data center energy efficiency metric. This energy efficiency metric (Himanshu 2014).
is defined by a formula PUE = Total Facility Power/Total Equipment Power. Where,
PUE= Power Use Elfectiveness.
A PUE value (David Cole 2011) of I would represent the optimal data center efficiency. In
practical, a PUE value of I means that all power going into the data center is being used to
power IT equipment. Anything above a value of 1 means there is data center overhead
required to support the IT load.
i. Duta Center l{etwork
Data center network (DCN) (Bin Wanga 2015) includes data center and provides the
connections to the data center, which is described by its network topology,
routing/switching equipment, and the protocols it uses.
DCN offers many features to help organize cloud computing for the following reasons (Bin
Wanga 2015):
o 'DCN permits the connection of thousands of data center servers in an efficient way, so
that cioud computing could simply expand its service by following the DCN topology.
o DCN offers traffic reliability and efficiency to massive machine-to-machine
communications in which activities from cloud computing emerge as the workloads
distributed on data center servers.
o DCN supports various virtualization techniques that help DCN to create Virtual
Machine (VM), virtual network, and virtual function. DCN should possess the
scalability to provide isolation and migration to massive numbers of virtual instances.'
\ \ ,/ \&' Y././,/ x
Data centerArchitecture overview . i:"i.
At present, 3 tier National Data Centre (NDC) is running under the ICT division. It's
storage capacity is I petabyte (One petabyte is equivalent to 10 lakh gigabytes.). The
storage capacity is already full. This NDC is going to be upgraded to tier 4 and will be
located in the Bangabandhu Hi-Tech City at Kaliakoir in Gazipur (The Daily Star 2018).
Chinese vendor ZTE Corporation is the developer ol this project. Uptime Institute (Data
Center certification provider in USA), the global authority on data center best practices.
will carry out a technicai audit of Bangladesh 4 tier NDC. This will be the first Tier 4
national data center, comprising the most advanced technology in use around the world as
part ol the government's move to boost the country's data storage capacity. Government is
trying to augment the capacity to up to 50 petabytes to store the huge data the country is
producing now to achieve Digital Bangladesh. Tier 4 data center wiil have 0.8 hours ol
interruption a year, which is haif of the disruption faced by a Tier 3 center.
7.1 NDC Service Packages
Following services are provided by Bangladesh NDC (BCC web portal20l9). Services are:
r' Hosting Services
o Web Hosting Service (Linux)
o Web Hosting Service (Microsoft)
o Application Hosting Services (Linux)
o Application Hosting Services (Microsoft)
/ Conrputing Services
o Load Balancer
/ hmallservice
o Zimbra Email
/ Database Service (Shared)
o Oracle
o Microsoft SQL Server
/ Storage Services
/ Collocation Services
o Rack Unit Allocation Services
/ Network Services
A global survey report on Data Centers have been published (Uptime Institute 201 8) by Data
Center certificate provider Uptime Institute. Some key challenges and complexities facing
data center managers globally were findings of that survey. Key findings are stated bellow.
' Efficient but more complex: PUE across the industry has improved, signaling an
improvement in data center efficiency. Today, operators are grappling with new
challenges, including the business case and cost ofhybrid IT approaches.
' Edge computing is coming: Operators are expecting to deploy significant new edge
computing capacity, which will add a layer of operational and management complexity.
\ \ \\,/ .w.,/
!.r:,.i I
Dovelopment Review'
Vohrme 28. 2019
. Rack density issues are growing: The highest rack densities reported at enterprise
and service provider data centers suggests that many operators face cooling challenges'
. DCIM is now mainstream: A small majority of data centers now have some type of
DCIM, and typically their impiementation has been successful (contrary to widespread
industry reports).
. Many data centers are unprepared for. or their managers are not expecting to be
affected, by climate change: Despite being vulnerable to increased temperatures, water
shortages, and extreme and sometimes disastrous weather events, most data center
operatJrs have determined that they either won't be impacted or are ignoring the problem.
. Data center skill shortages will intensify: ln this aging and overwhelming male
sector. most operato.r ur. rtrrggling with staffing issues. Yet most do not believe a lack
ofdiversity in their ranks is an issue to be concerned about.
Though data center investment is high at initial period but actually it is not so costly for
long ierm perspective. A data centeis has its huge impact in economic development and
orginizationai efficiency improvement. Given the large investments in data centers.
orlanizations should plin their data centers properly consider their available budget and
toGrrrr., for failures incl disruptions. The organization must anticipate its immediate and
long-term business needs and aiign them with its IT strategy. Data center Tier 4 level lnust
be Jhosen as standard when trying to establish sustainable and scalable data center. Policy
makers can look lorward to establish tier 5 Data Center in Bangladesh in futurc.
Ajay Ahuja (2015). State Data Center: Top Ten Design Parameters, Towards Next
Generation E-Government.
Beth Whitehead, Deborah Andrews, Amip Shah & Graeme Maidment (2014).'Assessing
the environmental impact of data centres part 1: Background, energy use and metrics',
Building and Environment 82 (2014) 151el59. DOI: 10.1016/j.buildenv.2014.08.021
Bin Wanga, Zhengwei . Ruhui Maa, Haibing Guana, Athanasios & Vasilakos (2015). A
survey oi dutu cJnter networking for cloud computing, Computer Networks 91 (2015)
528 541, Page 1.
David Cole (2011). Data Center Energy Efficiency - Looking Beyond PUE, No Limits
Software White Paper #4,Page 3.
Himanshu Gupta, Navid Hamedazimi & Samir Ranjan (2014).'Reconfigurable wireless
data center neiwork using free-space optics', Resesrch Foundution of State Universiqt of
It{ew York. US 1 4934863.
https://gigaom.coml2}l2ll0/15/4-types-of-data-centers/, Last visited on Novembet 23,2018., Last visited on Novembet 23, 2018., Last visited
on November 23.2018.
sn.UfatClfQ.dpbs, Last visited on November 23,2018.
https :// (20 I 1).Environmental Protection for Large Data Centers,
\ . \,/,/ ,/ ,/ 'x
... Data center is a physical environment facility intended for housing computer systems and associated components [2]. It means the data center is the core for the distributions and flows of all data and information in the organizations. ...
Full-text available
Bank XYZ is a banking company that has a Data Center department for the core of data and information distributions. Data Center department supports data storage, and providing the platform for the deployment of business applications. But some downtime issues still could emerge and require human resources to solve those issues. In this paper, we define some major issues those are connection down, server down, and broken data in UAT environment. Therefore, Bank XYZ’s data center needs an automation system for accurate and immediate solution. Automation system will need information from the log management application to do responses. Thus, this research’s objective is to have an automation design for operating automation in the Data Center department based on rules in log management. Research data are obtained from observation and interview methods. This research has discovered an automation design to help the Data Center department for helping them fixing their downtime issues. © 2020, World Academy of Research in Science and Engineering. All rights reserved.
Full-text available
Data centres consume high levels of energy to power the IT equipment contained within them, and extract the heat they produce. Because of the industry's heavy reliance on power, data centre metrics have historically used operational efficiency as a proxy for sustainability. More recently the industry has begun to recognise that its focus needs to go beyond energy consumption, with the creation of metrics for issues such as carbon, water and compute efficiency. However, single-issue metrics often consider only the operational phase, omitting impacts from other issues, during other stages in a facility's lifetime. Further approaches exist to assess more holistically the impact of data centres, such as building environmental assessment methods, but none have the capacity to capture fully the interlinked nature of a system, where improvements in one area and to one impact, can adversely affect a totally different area and totally different impacts. The following review of literature summarises the approach of the data centre industry to environmental impact, and provides direction for future research. Part 1 describes the energy consumption of the ICT industry and in particular data centres; current knowledge on the environmental impact of the industry; and how single-issue metrics have risen to prominence.
Full-text available
State Data Centers (SDC) are one of the key elements of the e-Governance Initiatives of the Government Of India. All the major States across the country will be hosting Citizen Services out of these State Data Centers. Many States are in the conception stage of designing and architecting these Data Centers. Many factors and design parameters need to be evaluated and considered while designing theses Data Centers. These design parameters are critical for the successful implementation of these State Data Centers. This paper discusses Top Ten, Most important Design parameters, which should be considered while designing State Data Centers.
A survey oi dutu cJnter networking for cloud computing
  • Zhengwei Ruhui Bin Wanga
  • Haibing Maa
  • Guana
Bin Wanga, Zhengwei. Ruhui Maa, Haibing Guana, Athanasios & Vasilakos (2015). A survey oi dutu cJnter networking for cloud computing, Computer Networks 91 (2015)
Environmental Protection for Large Data Centers
  • Itwatchdogs (20 I 1).Environmental Protection for Large Data Centers, \. \,/,/,/,/ 'x
Fire protection in data centers
  • Ltd Siemens Switzerland
Siemens Switzerland Ltd, Building Technologies Division (2015). Fire protection in data centers, BT_0074_EN, Page 7-10.
Planning and Designing Data Centres
  • Yee Keen Seng
  • Wu Xianghua
  • Lim Hwee Kwang
Yee Keen Seng, Wu Xianghua, Lim Hwee Kwang, Wong Marn Chee & Ang Choon Keat. 'Planning and Designing Data Centres', Data Horizons, urce/dsta-aboutiplanning-and-designing-data-centres.pdf?sfvrsn=2,Page 102-103
Uptime Institute Global Data Center Survey Report
Uptime Institute (2018). "Uptime Institute Global Data Center Survey Report". https://da 811112018-data-center-industry-survey.pdf, Acessed on 3rd }i4.ay 2019.