3D Vizualization of Large Scale Data Centres

Giannis Drossis

, Chryssi Birliraki

, Nikolaos Patsiouras

, George Margetis

and Constantine Stephanidis

1,2

Institute of Computer Science, Foundation for Research and Technology Hellas,

N. Plastira 100, Vassilika Vouton, GR-70013, Heraklion, Crete, Greece

Computer Science Department, University of Crete, Heraklion, Crete, Greece

Keywords: Data Centre Visualization, Big Data Visualization, Interactive Real-Time Visualization, 3D Visualization,

Virtual Reality, Immersive Visualization, Data Centre Infrastructure Management,

Large High-Resolution Displays, Data Centre Resource Planning, Gestural Interaction.

Abstract: This paper reports on ongoing work regarding interactive 3D visualization of large scale data centres in the

context of Big Data and data centre infrastructure management. The proposed approach renders a virtual area

of real data centres preserving the actual arrangement of their servers and visualizes their current state while

it notifies users for potential server anomalies. The visualization includes several condition indicators, updated

in real time, as well as a color-coding scheme for the current servers’ condition referring to a scale from

normal to critical. Furthermore, the system supports on demand exploration of an individual server providing

detailed information about its condition, for a specific timespan, combining historical analysis of previous

values and the prediction of potential future state. Additionally, natural interaction through hand-gestures is

supported for 3D navigation and item selection, based on a computer-vision approach.

1 INTRODUCTION

Big data analysis is at the centre of modern science

and business (Eaton et al. 2012). On the other hand,

the need for additional storage space due to the

massive growth of the data volume (Schneider 2012)

has resulted into the expansion of existing data

centres.

High quality and reliable services are essential for

the data centres. Any potential issues that may arise

should be dealt with as soon as possible in order to

eliminate, or at least minimize, the downtime of

servers. The increasing dependency on the reliability

and responsiveness of systems and services requires

continuous and real-time monitoring, creating the

need for real-time rich interactive visualizations that

assist data centre experts.

Additionally, several approaches successfully

manage to not only identify problems, but also to

reason on the cause of the issues and predict future

behaviours. Such examples include Root Cause

Analysis, comprising many different tools, processes

and philosophies, and Data Centre Predictive

Modelling (Moore, Chase and Ranganathan, 2006),

forecasting the performance of a data centre into the

future respectively.

Data centre infrastructure management consists

an integral part of Big Data, as this sector provides

the means for supporting Big Data storage, retrieval

and monitoring as a whole. Although much work

exists in literature regarding both Big Data

visualization (LaValle et al., 2011) and (Keim et al.,

2013) and data centre infrastructure management

(Moore et al., 2005), (Moore et al.,2006) and (Harris

and Genq, 2015), limited research targets the

direction of visualizing the state of the data centres.

Several indicators exist that can provide insights of

the current state of each server, such as CPU usage,

power consumption, network load, etc. A major issue

for data centre infrastructure management involves

real-time monitoring of the existing facilities. In this

context, this paper addresses the need for an intuitive,

rich and adaptable visualization of all the hardware

and software components that exist in a data centre.

Maintenance of Big Data infrastructure involves

providing continuous monitoring and troubleshooting

any potential issues that may arise (such as hardware

inefficiencies) as soon as possible. The growth of data

volumes results into the further expansion of data

centres, and consequently, the addition of further data

servers. Potential issues may either be resolved

remotely or physically on site; in any case, the first

388

Drossis, G., Birliraki, C., Patsiouras, N., Margetis, G. and Stephanidis, C.

3D Vizualization of Large Scale Data Centres.

In Proceedings of the 6th International Conference on Cloud Computing and Services Science (CLOSER 2016) - Volume 1, pages 388-395

ISBN: 978-989-758-182-3

step of solving a problem is identifying it. This fact

adds up to the complexity of locating the area of

interest instantly, a process which is solved within

minutes only in 26% of the cases (Cole, 2012).

The proposed approach aims at creating an

adaptive, multi-purpose system which is able to allow

database experts to manage a data centre effectively

and efficiently. To this end, the system employs a

three-dimensional virtual environment to visualize

the data centre and supports gesture-based

interaction. The innovative aspects that enhance the

visualization of the data centre layout include the

notification of problematic units, the on-demand

exploration of server details and the historical

overview of past values.

Furthermore, this approach aims at generating a

self-explanatory environment which will minimize

the need for training users. To achieve this goal, the

proposed approach focuses on providing a virtual

environment of large-scale data centres that is very

close to the real ones, realising in this way procedures

for identifying and addressing a problem similar to

those procedures that are followed by the experts on

site. Moreover, it introduces alternative ways of

interaction, based on hand gestures, facilitating thus

different scales of installations, from a desktop

monitor to the very large screens of a situation room.

In this context, the proposed work aims to fill-in

the gap of innovative real-time data centre

infrastructure visualization and management in the

context of Big Data. Furthermore, the method

combines state-of-the-art 3D visualization techniques

and gestural interaction. In the following sections,

this paper reports on the work presented in literature

in the related areas of research, presents the proposed

approach in terms of system design, interaction and

data storage/retrieval, gesture-based interaction and

finally outlines conclusions and potential future

directions of interest.

2 RELATED WORK

Data centre infrastructure management heavily

involves finding the optimal architecture by

identifying imbalances in the existing setups. The

work presented by (Moore et al. 2005) analyses

existing facilities’ performance data, as well as

workload playback methods. Furthermore, data

centres’ impact to the environment is a subject of

research in terms of energy consumption (Andrae and

Edler, 2015), workload distribution and isolation

(Sfakianakis et al., 2014) and data centre network

virtualization.

3D visualizations exist in various forms in

literature due to their ability to display increased

information at a glance. The authors of (Reda et al.,

2013) apply 3D visualization in hybrid-reality

environments in order to visualize large

heterogeneous data.

Regarding real space visualization, literature

mainly contains work focusing on dissimilar

simulated areas, such as architecture (Koutamanis,

2000), homes (Lozada and De la Rosa, 2014) and

(Lazovik et al., 2009), and industrial setups (Lawler,

2010) and (Chen, 2012). Lawler focuses on creating

a toolkit to improve performance using a packaging

line as a case study, while Chen emphasises on floor

area reduction and visual management. The technique

of color-coding is widely adopted in literature (Ware,

2012) and especially in the case of visualizing

complex information which can be categorised

(Gölitz et al., 2013) and (Kriglstein et al., 2013).

As far as gestural interaction is concerned, hand

gesturing is an approach widely adopted in literature

as a straightforward, intuitive and easy to memorise

interaction method. According to Aggarwal

(Aggarwal and Ryoo, 2011) gestures are defined as

elementary movements of a person’s body part, most

commonly involving the hands, arms, face, head,

defining the expressive atomic components that

describe the meaningful motion of a person. Gestural

interaction is both applied in the air (Drossis et al.,

2013) and (Ren et al., 2013) and using touch (Valdes

et al., 2014). Computer vision based hand gesture

tracking allows interaction in a similar manner to

every day human to human communication without

the burden of actually coming to physical contact

with any object. Despite the considerable research

efforts and advances, hand-based gesture recognition

and incorporation in interaction remains a

challenging task.

The presented work aims to address new aspects

that have emerged with the vast inflation of data

centres, as a result of the need for very large scale

infrastructures accommodating big data and cloud

services. To this end, the proposed approach extends

existing approaches for data centres’ monitoring, by

providing a more realistic visualization combined

with natural interaction.

3 SYSTEM DESIGN

The Data Centre 3D Visualization application aims at

helping data centre experts to get an intuitive

overview of a specific data centre room, regarding its

current condition. Additionally, the application

3D Vizualization of Large Scale Data Centres

389

facilitates inspection of the racks and servers by the

users and warns them, in an intuitive manner, about

situations that need further investigation, such as an

anomaly regarding a particular set of servers that may

bring to surface malfunctions or degraded operation.

Taking into account the various potential contexts of

use, ranging from an office to a control room, the

application is able to be deployed in a laptop or a PC,

featuring interaction with the mouse, as well as in

large displays, with gestures.

Figure 1: Main Scene view.

3.1 3D Overview

The main screen of the Data Centre 3D Visualization

application, which comprises a virtual representation

of a data centre room and the basic interactive UI

components (e.g., navigation buttons) is depicted in

Figure 1. All the servers of the room are grouped and

displayed as 3D racks according to their physical

location in space. Each rack may contain at most 40

units, each displayed as a slice with a specific colour,

annotating its current condition [Figure 6Figure 3].

Overall, the virtual data centre room is constructed as

a grid in the 3D space. The virtual environment that

encloses the scene is spherical and the servers’ grid is

placed in the centre, so that users can have a 360

degrees overview.

The application provides details and insights of

the displayed data centre room through its intuitive

user interface. In more details, at the left side of the

application’s viewport a menu of five components is

displayed, highlighting elementary values of the data

centre’s servers such as temperature, CPU load,

power consumption, network Bytes In/Out, Disk I/O

[Figure 1]. Upon the selection of one of these values,

the visualized information of the data centre is

updated accordingly, representing the current status

for the selected characteristic. This way, the users are

able to instantly switch between different server

characteristics and correlate the displayed state of the

data centre with specific server characteristics (e.g.,

CPU Load and Power Consumption: usually, a high

CPU Load is expected to result into a higher power

consumption).

Furthermore, the application provides historic

information of the data centre state by allowing users

to select the date and time of the information that is

visualized. Upon the selection of a time frame, all the

servers update their values accordingly. If the

selection is performed in the close-up view of a rack,

then additional options are provided in order to

present the history of the selected unit [Figure 3].

Figure 2: [Left] Visualization Μenu Οptions, [Middle-Up]

Scene Sparseness Filtering, [Middle-Down] Date Time

Filtering, [Right] Displayed Unit Colours Filtering.

In addition, the system offers filtering and

arrangement options [Figure 2], allowing users to

manipulate the information provided and focus on

specific aspects of the data centre. In more details,

filtering according to the state of the displayed servers

(color-coded) allows exclusion of elements which are

not in the field of the user’s interest, resulting in a

comprehensible and clear view of the available

information (e.g., the user can easily view all the

servers in critical state). On the other hand,

arrangement according to the level of sparseness of

Figure 3: Close-up state shows the units of the selected rack

and focuses on the "worst" one, with number 33. On the left

the UI-information for this unit and on the right the line

charts for the specified history.

DataDiversityConvergence 2016 - Workshop on Towards Convergence of Big Data, SQL, NoSQL, NewSQL, Data streaming/CEP, OLTP

and OLAP

390

the data centre, aids in defining the view of the data

centre (e.g., the servers can be densely placed in order

to provide an overview of the data centre or sparsely

so as to facilitate inspection of server values in detail

over specific areas of the data centre room).

3.2 Navigation

Navigation in the scene is accomplished through an

orbit virtual camera providing, in this way, different

levels of details. For example, if the camera moves

forward then the scene focuses towards a particular

server (zoom in) and the user gets a closer and

detailed view of the server details and its units. On the

other hand, if the camera moves away (zoom out)

then the user gets the whole picture of the data centre.

Furthermore, the camera can also move left and right

to provide a better view of all the sides of the data

centre. Upon the selection of a specific rack, the

camera zooms into it, allows moving up or down to

focus on a specific unit, and explore the provided

information.

3.3 Close-up View

Investigation of a rack is achieved by either using the

controls at the bottom of the screen or performing

appropriate hand gestures. This view [Figure 3] is

used for inspecting information on a per-server level,

displaying the current values of the server’s attributes

(e.g., temperature, CPU load, etc.) along with line

charts. Any of these charts can be enlarged on users’

request, in order to provide a more detailed view. The

close-up view of a rack provides up-down navigation

to the user for selecting a particular unit. When the

user does so, the current values of the selected unit

are illustrated on the left side of the screen, while

history values are depicted as line charts on the right

side. On top of the screen, there are two sliders: the

displayed timestamp slider and the time window

slider. The first slider provides the ability to the user

to select a past time value for the selected unit in order

to investigate its previous states. The time window

slide sets the size of the time frame used for the

calculation of the visualization of the mean values

illustrated by the charts. For instance, if a user

increases the value of the time window slider then

he/she increases the timespan over the means of

previous unit’s values are estimated and displayed,

getting in this way a broader view of the unit’s history.

3.4 Anomaly Detection Visualization

The purpose of the 3D data centre visualization is to

provide a real-time view of a data centre and to

facilitate the efficient handling of a critical situation

when it is turned up. Therefore, the system provides

a mechanism that supports the representation of

anomaly detections regarding the temperature or

power consumption. When a problematic unit is

detected, then the colour of the rack that contains this

unit changes to red and a notification message pops-

up above it, displaying a short description of the

problem and the unit’s number [Figure 4].

Figure 4: Upon anomaly detection, the server has changes

colour accordingly (red).

3.5 Interaction Techniques

The application supports mouse-based and gesture-

based interaction, aiming to address the potential

contexts of use, i.e., a typical desktop environment

and a control room.

3.5.1 Mouse-based Interaction

The interactive elements of the scene include both the

server racks and the server units. Upon mouse

hovering a server rack, a notification pops-up

presenting basic information (i.e., the rack’s id, the

room it is located in, and its coordinates in the room,

if the unit is in critical state). If the user clicks on a

specific rack, the scene changes to a close-up view,

the camera zooms in towards the selected rack and

specifically displays the unit of the rack which is in

the “worst situation”, and which defines the criticality

level of the server [Figure 3]. While in close-up view,

the user can select a specific unit of a rack, through a

mouse click. Then, at the right side of the rack, four

different line charts are displayed, illustrating the

history of the selected unit, as already discussed in

section “Close-up View”. If the user hovers the

mouse pointer over a specific chart, then the chart

expands in order to display a better view of the unit’s

history and its temporal evolution [Figure 5].

3D Vizualization of Large Scale Data Centres

391

Figure 5: Maximized chart on mouse over the interactive

item at the right screen side.

In addition, aiming to facilitate users’ navigation in

the 3D environment, a navigation tool has been

developed, through which the view of the camera is

controlled [Figure 6]. This component comprises four

navigation buttons for moving the camera forward,

backward, left and right, as well as a minus and a plus

button for camera zoom out and in accordingly. These

buttons are triggered on hover. For example, as long as

the mouse pointer hovers the up arrow button the

camera moves forward. The navigation component

changes when a rack is selected and the scene switches

to the close-up view. In this state, the navigation

buttons provide only the up or down functionality since

the server’s units are stacked vertically on the rack.

Furthermore, there is an exit button in order allowing

users to return to the full scene view.

Figure 6: Camera navigation tools.

3.5.2 Gesture-based Interaction

In order to make the interaction more natural and

friendly to the user, especially when the system is

deployed in a large-screen setting, gesture-based

interaction is supported. The current version supports

a small vocabulary of gestures that enable users’

interaction with the 3D scene using a cursor metaphor

(hand-mouse).

In more details, through a hand tracking

mechanism, the user’s movements are filtered and a

mouse event emulator is raised and listens to the

incoming events that correspond to the user’s moves.

As a result, when the user moves his right hand in the

space in front of him, the mouse moves accordingly

in the scene [Figure 7]. In order to select an

interactive element the user has to raise his left hand

and open the palm, showing at the same time with his

right hand the component that he wants to select.

Figure 7: Gesture-based interaction.

Future versions of the application will support

gesture-based interaction through a larger vocabulary

of intuitive gestures (e.g., move hands apart for

zooming in). Nevertheless, the hand-mouse metaphor

will continue to be supported, as it is an easy-to-learn

interaction modality for first-time users of gesture-

based interaction.

4 DATA STORAGE AND

RETRIEVAL

4.1 Input Data and Analysis

In order to create the scene layout and fill in the server

racks and the corresponding units with their values, a

mechanism has been implemented to parse an input

file (.csv format) and to retrieve all the data written in

it, analyse it - if needed - and create the necessary data

structures in memory for information representation

of the scene data elements.

Figure 8: Input CSV file format.

Each line of the file features comma-separated

values providing information for a specific unit of a

server of a room in a specific Date and Time. The

comma-separated values represent information

regarding the temperature of the server, its CPU load,

power consumption, network bytes in/out and Disk

I/O. The parsing of the file is based on the specific

format of each line, as shown in Figure 8. For

DataDiversityConvergence 2016 - Workshop on Towards Convergence of Big Data, SQL, NoSQL, NewSQL, Data streaming/CEP, OLTP

and OLAP

392

example, “1410768000, RA_A24_8, 0.83,48180,

34080, 810180, 64.6, 344” reports that:

 1410768000 is the Date Time (long)

 RA_A24_8 the information refers to Room A,

for the server which is located in Row A,

Column 24 and the unit for which the data in

this line refer to is the 8th.

 0.83 is the CPU load in the specific timestamp

(float)

 48180 are the Network Bytes In for the

specific timestamp (long)

 34080 are the Network Bytes Out for the

specific timestamp (long)

 810180 is the Disk Input/Output for the

specific timestamp (long)

 64.6 is the temperature load for the specific

timestamp (float)

 344 is the power consumption for the specific

timestamp (integer)

The data provided in the input file is used to

populate the data structures that are required to

visually represent the aforementioned information in

the 3D scene. In more details, the following data

structures are used: (a) a server-description structure,

featuring the room in which the server is stored, the

position of the server in the grid layout, and the units

of the server, (b) a unit-description structure, which

involves all the unit values that have been explained

above (CPU load, network bytes in, network bytes

out, disk I/O, temperature, power consumption), as

well as the timestamp information.

4.2 Scene Creation

Apart from the data structures needed, a scene

generator has been created in order to visualize the

data according to the information stored in the file.

The scene generator asynchronously creates all the

3D racks with their units and all the coloured values

in them and places them according to their positions

given in the input file for the specific room. As

illustrated in Figure 1 and explained in section “Input

Data and Analysis”, positions are defined by rows and

columns, which is an easy to understand and effective

way to visually map them in the scene as a grid with

the corresponding rows and columns.

4.3 Real-time Communication

Information regarding the servers’ state is acquired in

real time, both in terms of raw data (such as

temperature, CPU usage, etc.) and in terms of

anomalies. Information retrieval is accomplished by

automatic anomaly detection and root cause analysis

CEP Operators, which is part of a system developed

in the context of the LeanBigData FP7 project

[LeanBigData, 2014-2017]. The data currently

provided comes from recorded datasets, while

ongoing work involves the provision of real-time

information. The communication is implemented

using REST API endpoints and provides both current

and previous values for any given server.

5 GESTURE-BASED

INTERACTION

For supporting gesture based interaction a framework

has been implemented based on the approach

discussed in (Michel et al., 2014). The adopted

framework encompasses a collection of techniques

that enable robust, real-time and efficient gesture

recognition based on visual information acquired by

an RGBD camera. The framework encompasses hand

detection, hand tracking, hand posture recognition

and hand gesture recognition processes and interprets

the recognized hand gestures to windows native

mouse events, acting as a Human Interface Device

(HID). To this end, it can be installed on any

computer with windows operating system, providing,

in a transparent way, gesture – based interaction to

any windows application.

The employed gestural vocabulary is composed of

three different classes of physical hand-arm actions.

Mouse Button Press: The case of a mouse button

press is recognised if the posture of a tracked hand

changes from “closed hand” to “open hand” and the

“open hand” posture is maintained for a number of

consecutive frames.

Mouse Button Release: The case of a mouse

button release is recognized if the posture of a tracked

hand changes from “open hand” to “close hand” and

the “closed hand” posture is maintained for a number

of consecutive frames.

Cursor Move: A hand has a posture that is

classified as “index up”. Then, the 2D projection of

its centroid in the image plane as it moves around,

defines the x-y coordinates of the mouse cursor in the

application area.

6 CONCLUSIONS AND FUTURE

DIRECTIONS

The work presented in this paper attempts to combine

innovative, state-of-the-art technologies in the fields

3D Vizualization of Large Scale Data Centres

393

of Big Data anomaly detection, information

visualization and computer vision gesture

recognition, in order to deal with visualization needs

for Big Data and data centre infrastructure

management.

The proposed approach primarily deals with the

monitoring and the intuitive display of existing data

centres’ information, using their actual layout, in

order to inform data centre experts about the servers’

current state and assist navigation in actual space. The

proposed approach takes advantage of 3D rendering,

providing seamless transition from the data centre’s

overview to on-demand specific server information.

Finally, the presented work is designed not only to

suit traditional desktop interaction but also to support

natural interaction by employing gesture-based

interaction.

Future work involves enriching the gestural

vocabulary and conducting an in-depth qualitative

and quantitative evaluation, in order to assess the

system’s usability, scalability and the overall user

experience. Another challenging issue upon which

further research can be directed is the ability to

incorporate the visualization of relationships between

servers in the system.

Finally, this work aims to act as a starting point

for developing a complete framework for Big Data

Infrastructure Management. Due to the nature of Big

Data, a plethora of information exists that is

significant and meaningful for data centre experts,

constituting a very demanding area in the

interdisciplinary domain of 3D Graphics, Human-

Computer Interaction and Visual Big Data Analytics.

ACKNOWLEDGEMENTS

This research has been partially funded by the

European Commission under project LeanBigData

(FP7-619606)

REFERENCES

Aggarwal, J., Ryoo, M., 2011. Human activity analysis: A

review. ACM Comput. Surveys. 43.

Andrae, A. S., & Edler, T. (2015). On Global Electricity

Usage of Communication Technology: Trends to

2030. Challenges, 6(1), 117-157.

Chen, C. P., & Zhang, C. Y. (2014). Data-intensive

applications, challenges, techniques and technologies:

A survey on Big Data. Information Sciences, 275, 314-

347.

Chen, Z. (2012). Production system improvement: floor

area reduction and visual management (Doctoral

dissertation, Massachusetts Institute of Technology).

Cole, D. (2012). Data center infrastructure

management. Data Center Knowledge.

Drossis, G., Grammenos, D., Birliraki, C., & Stephanidis,

C. (2013). MAGIC: Developing a Multimedia Gallery

Supporting mid-Air Gesture-Based Interaction and

Control. In HCI International 2013-Posters’ Extended

Abstracts (pp. 303-307). Springer Berlin Heidelberg.

Eaton C., Deroos D., Deutsch T., Lapis G. and Zikopoulos

P. (2012), Understanding Big Data: Analytics for

Enterprise Class Hadoop and Streaming Data.

Harris, M., & Geng, H. (2015). Data Center Infrastructure

Management. Data Center Handbook, 601-618.

Gölitz, P., Struffert, T., Lücking, H., Rösch, J., Knossalla,

F., Ganslandt, O., & Doerfler, A. (2013). Parametric

color coding of digital subtraction angiography in the

evaluation of carotid cavernous fistulas. Clinical

neuroradiology, 23(2), 113-120.

Keim, D., Qu, H., & Ma, K. L. (2013). Big-data

visualization. Computer Graphics and Applications,

IEEE, 33(4), 20-21.

Koutamanis, A. (2000). Digital architectural

visualization. Automation in Construction, 9(4), 347-

360.

Kriglstein, S., Wallner, G., & Rinderle-Ma, S. (2013). A

visualization approach for difference analysis of

process models and instance traffic. InBusiness Process

Management (pp. 219-226). Springer Berlin

Heidelberg.

LaValle, S., Lesser, E., Shockley, R., Hopkins, M. S., &

Kruschwitz, N. (2011). Big data, analytics and the path

from insights to value. MIT sloan management

review, 52(2), 21.

Lawler, M. E. (2010). Improving shop floor visualization

and metrics (Doctoral dissertation, Massachusetts

Institute of Technology).

Lazovik, A., Kaldeli, E., Lazovik, E., & Aiello, M. (2009).

Planning in a smart home: visualization and simulation.

In Application Showcase Proceedings of the 19th

(ICAPS) Int. Conf. Automated Planning and

Scheduling.

LeanBigData - Ultra-Scalable and Ultra-Efficient

Integrated and Visual Big Data Analytics, Funded by

FP7 ICT 619606, 2014-2017, http://leanbigdata.eu/,

http://cordis.europa.eu/project/rcn/191643_en.html

(last accessed on 21/1/2016).

Lozada, C., & De la Rosa, R. (2014, June). Simulation

platform for domotic systems. In Communications and

Computing (COLCOM), 2014 IEEE Colombian

Conference on (pp. 1-6). IEEE.

Michel, D., Papoutsakis, K., & Argyros, A. A. (2014).

Gesture recognition supporting the interaction of

humans with socially assistive robots. In Advances in

Visual Computing (pp. 793-804). Springer

International Publishing.

Moore, J., Chase, J., Farkas, K., & Ranganathan, P. (2005,

February). Data center workload monitoring, analysis,

DataDiversityConvergence 2016 - Workshop on Towards Convergence of Big Data, SQL, NoSQL, NewSQL, Data streaming/CEP, OLTP

and OLAP

394

and emulation. In Eighth Workshop on Computer

Architecture Evaluation using Commercial Workloads.

Moore, J., Chase, J. S., & Ranganathan, P. (2006, June).

Weatherman: Automated, online and predictive thermal

mapping and management for data centers.

In Autonomic Computing, 2006. ICAC'06. IEEE

International Conference on (pp. 155-164). IEEE.

Reda, K., Febretti, A., Knoll, A., Aurisano, J., Leigh, J.,

Johnson, A., ... & Hereld, M. (2013). Visualizing large,

heterogeneous data in hybrid-reality

environments. IEEE Computer Graphics and

Applications, (4), 38-48.

Ren, Z., Yuan, J., Meng, J., & Zhang, Z. (2013). Robust

part-based hand gesture recognition using kinect

sensor. Multimedia, IEEE Transactions on, 15(5),

1110-1120.

Schneider, R.D. (2012), Hadoop for Dummies Special

Edition, John Wiley&Sons Canada.

Sfakianakis, Y., Mavridis, S., Papagiannis, A.,

Papageorgiou, S., Fountoulakis, M., Marazakis, M., &

Bilas, A. (2014, November). Vanguard: Increasing

Server Efficiency via Workload Isolation in the Storage

I/O Path. In Proceedings of the ACM Symposium on

Cloud Computing (pp. 1-13). ACM.

Valdes C., Eastman D., Grote C., Thatte S., Shaer O.,

Mazalek A., Ullmer B., Konkel M. (2014). Exploring

the design space of gestural interaction with active

tokens through user-defined gestures. ACM.

Ware, C. (2012). Information visualization: perception for

design. Elsevier.

3D Vizualization of Large Scale Data Centres

395