ENHANCING STATISTICS TEACHING WITH A VIRTUAL LAB

A Case Study of Seamless Local and Remote Computing

Araceli Gar

ın

Facultad de CC.EE. y Empresariales, University of the Basque Country, Bilbao, Spain

Fernando Tusell

Facultad de CC.EE. y Empresariales, University of the Basque Country, Bilbao, Spain

Aitziber Unzueta

Facultad de CC.EE. y Empresariales, University of the Basque Country, Bilbao, Spain

Keywords:

e-Learning, Remote computing, Virtual lab, Client-server.

Abstract:

Some Statistics and Data Analysis courses demand a fair amount of computing. While there are excellent free

source tools which can be given away to students, their seamless integration requires a fair amount of work and

is challenging to the less computer-savvy students. In the past we have addressed the problem by compiling

and integrating the necessary tools in CD-ROM’s and providing local computing facilities, but this has proved

impractical on a number of counts. Trial and error has led us to ﬁnding a setup with which we have ﬁnally

solved most problems.

1 INTRODUCTION

Teaching Statistics and related subjects such as

Econometrics or Operations Research to students of

Economics presents various challenges; not least, ap-

plied courses on these subjects require a fair amount

of practice, all of which entails the use of a computer.

In the old, pre-1990 days, we would provide ac-

cess to a computer lab, typically made of a mainframe

or minicomputer with terminals connected in its close

vicinity. Any computer work would be done there or,

rarely, from a distant location through a dial up line.

The advent and rapid dissemination of personal

computers changed all this. For over a decade, the

“PC room”, a bunch of PC’s possibly connected in

a local area network, was the dominant paradigm in

ours an many other schools.

In Section 2 we list the advantages and shortcom-

ings of both the old computer lab model and the PC

room-based instruction, as we see them. We also

explain our moves in trying to cope with perceived

shortcomings in our infrastructure. In Section 3 we

describe our experience on how a dual-faceted com-

puter Lab, ﬁtted with PC’s yet having a work-station

as its cornerstone, provides the best of both models,

increases the productivity of students and teachers,

and can be run with a (relatively) modest input of

skilled labour.

2 THE PAST

2.1 The Old-style Computer Lab

The setup described in Section 1, which was prevalent

at universities all over the world until the ﬁnal eighties

of the last century, was not without advantages. The

requirements of skilled labour were moderate: there

was a single machine, or only a few, to be run. Dumb

terminals, even X-terminals with graphical capabili-

ties, were an “install-and-forget” type of task, and run

until physically worn out.

There was no question of viruses spreading, no

questions of illegal copying, and much reduced con-

cerns about security: there was little an inexperienced

user could do to cause damage. Since dumb terminals

were of little use to most users, hardware theft was al-

518

Garín A., Tusell F. and Unzueta A..

ENHANCING STATISTICS TEACHING WITH A VIRTUAL LAB - A Case Study of Seamless Local and Remote Computing.

DOI: 10.5220/0003480905180522

In Proceedings of the 3rd International Conference on Computer Supported Education (UeL-2011), pages 518-522

ISBN: 978-989-8425-50-8

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

so a rarity.

Finally, from the teaching standpoint which is

what interests us here mostly, the physical proxim-

ity of students having the same problems afforded

many opportunities for cooperative learning and mu-

tual help.

The shortcomings, however, are also evident: es-

sentially all work had to be done in one place, and

users depended on computer staff for the simplest

tasks, such as bringing in new data whose size meant

that it could not be typed at a terminal.

2.2 The PC Room

The PC rooms changed all of this. Suddenly, the users

were masters of themselves. They found at the school

the same hardware, the same operating system and

software tools that they were used to work with at

home. There was no retraining, no learning curve,

no dependence on any one to move data, which could

travel in ﬂoppies or, later, pen drives.

Alas, this very ﬂexibility was not without incon-

venients. PC’s could be infected and their hard drives

erased. In order to have shared services such as stor-

age or printing, local area networks (LAN’s) had to

be set up. Linking different machines into the same

LAN, managing authentication in a centralized man-

ner, ensuring consistency and integrity of the software

installed, protecting users from themselves and each

other became tasks that dwarfed the effort previously

required for the administration of a single machine.

Since, in addition, hardware and software have

sort useful lives because of technical obsolescence,

managing computer rooms is a task extremely de-

manding of resources. When all is taken into consid-

eration, the advantage of standard, off-the shelf hard-

ware is negated by the complexity of the installation

and maintenance.

2.3 The “Help Yourself” Approach

The increase in availability and quality of free soft-

ware opened new perspectives: software could be

given away to students in a CD or DVD and, at

least for work requiring only moderate resources, they

were able to work at home or wherever they could

bring their laptops. It seemed for a while that provid-

ing computer resources to students was no longer a

problem.

It soon became evident that this was not the case.

For one thing, preparing, customizing and integrating

all the necessary tools is a lot of work —and a work

that needs to be redone frequently so that the software

stays reasonably current. On the other hand, installa-

tion in a variety of hardware, with different operating

systems or different versions of the same operating

system, requires much testing, is very error-prone and

in our case meant that a sizeable proportion of the stu-

dents failed to have working installations.

Another drawback is that commonly used and

voluminous information (like digital cartography or

time series data collections) is difﬁcult to share and

keep current by way of handing over CD-ROM’s.

From the point of view of learning, this approach

also meant individual work, with greatly diminished

opportunities for interaction among students —a fac-

tor whose importance cannot be overemphasized, as

in our experience they learn much more easily applied

skills by interacting among themselves than in isola-

tion.

3 THE VIRTUAL LAB

3.1 Hardware and Software Setup

Around 1999 it became obvious that none of the ap-

proaches we had tried had been fully, or even mod-

erately, successful. In an attempt to regain the ad-

vantages of the old-style computer lab and keep the

advantages of modern PC’s, we decided to mix both;

the success has been above our expectations, and we

think this success is the outcome of a delicate interac-

tion of several factors, which we did not quite foresee.

We secured the premises and asked for the fund-

ing of a new lab, named Laboratory of Quantitative

Economics (LQE) after its intended users, graduate

students of said specialty. The design goals were:

1. It should provide a place for interaction among

students. Thus, each of them would have his or

her own desk and personal computer and there

would be some facilities to be shared, like a

printer, and an area for socializing.

2. It would support work in Statistics and Econo-

metrics, and target areas that we felt in need of

a boost, like Spatial Statistics and Data Mining.

3. It would have to be run on a very low overhead,

without requiring dedicated staff.

4. It would be based on free software, so students

willing and able to do so could replicate whatever

they found useful in their private machines.

The layout of the Laboratory is completely stan-

dard and can be seen in Figure 1. A machine is acting

as a server, providing among others authentication,

ﬁle storage and printing services. A number of PC’s

(we usually have between one and two dozen machi-

ENHANCING STATISTICS TEACHING WITH A VIRTUAL LAB - A Case Study of Seamless Local and Remote

Computing

519

nes running) are networked in the same Ethernet bus,

currently 100Mbit Ethernet. Everything is connected

to the Internet.

Internet

Server

Figure 1: Sketch of arrangement at the Laboratory for

Quantitative Economics.

The server is a 64bit machine running Linux; we

use Debian

, and are satisﬁed with it, but a number of

other distributions offer also 64 and 32 bit versions of

Linux. What is essential, as we will discuss below, is

that the same software is available in the server and

the client machines.

Students have each his own desk and PC. All PC’s

are fully autonomous machines running 32 bit Debian

Linux. However, user ﬁles are stored in the server; the

PC’s mount the relevant directory so it appears to be

local; this is transparent to the user.

Software installed in both the server and PC’s in-

clude editors and word processing tools (we encour-

age the use of Emacs, (Stallman, 1997), and L

(Lamport, 1994), but users are free to use the Open

Ofﬁce suit or other tools), statistical and econometric

software (R, (R Development Core Team, 2008), and

Gretel

, among others), GIS tools (Qgis

, GRASS

database tools (PostgreSQL

, PostGIS

), graphics

programs like Gimp

, and an assortment of oﬁmatics

and productivity tools.

The important point is that the same software

(with very few exceptions, which are not available or

are only available in a different version for 32 and 64

bit machines) is installed in the server and PC’s. This

brings about two beneﬁts:

• Users can at any point log to the server to start

jobs too large for their PC’s. The server is a mul-

ticore machine each of whose cores runs typically

3 or 4 times faster than a single PC. Since all user

ﬁles reside in the server, no ﬁle movements are

involved.

• More importantly, each user is able to log re-

motely and see exactly the same environment he

See (Krafft, 2005) or http://www.debian.org.

See http://gretl.sourceforge.net

See http://www.qgis.org/.

See http://grass.fbk.eu or (Neteler and Mitasova, 2007).

Described in http://www.postgresql.org and also in

(Momjian, 2001).

See http://postgis.refractions.net/.

See http://www.gimp.org.

or she would see in the Lab. All the software and

ﬁles are there: there is no need to carry anything

in pen drives, to upload or download ﬁles.

For the second beneﬁt to be fully realized, it is im-

portant that they have suitable means at home. Nowa-

days, personal machines are ubiquitous, and almost

every student has one; most have also high speed In-

ternet access.

While the above setup provides a feasible work-

ing environment, we did not reap the full beneﬁts un-

til we added a last and very important piece of soft-

ware. There are a number of X-terminal software em-

ulators allowing a remote machine to serve as an X-

terminal. However, the protocol is fairly verbose, and

the amount of information to be transferred fairly vo-

luminous: every keystroke or click requires process-

ing on both the local and remote machine. In our ex-

perience, this leads to latency times which are unac-

ceptable for interactive work. Response is much too

sluggish when graphical applications are used.

On the other hand, installing these emulators at

home (usually on machines running various versions

of Windows) was a hurdle for the less computer-savvy

users, a problem compounded with the fact that addi-

tional layers of software are necessary for the under-

lying Secure Shell (ssh) protocol.

This problem has been much alleviated by the use

of software which compress and streamline the X pro-

tocol. Several such pieces of software are in existence

using the so-called NX technology

. After some ex-

perimentation, we have settled for the freely available

(although not free source) implementation of NoMa-

chine

With this emulator installed, interactive work be-

comes almost as fast in a remote machine as it is lo-

cally. The only problem we have found is that X-

terminal emulation is impractical or even impossible

with some notebooks supporting only low screen res-

olution modes.

3.2 Administration

It may appear that the setup described shares the prob-

lems associated with PC rooms. This is not the case.

It is true that each PC has to be installed, but the pro-

cess can be automatized to a large degree

, especially

if you have control over the network and can have

the machines boot from a remote image. Even if you

can’t, the process of inserting a CD-ROM on each PC

For a description and overview of alternatives, see

http://en.wikipedia.org/wiki/NX technology.

Available at http://www.nomachine.com.

See for instance FAI, Fully Automatic Installation,

http://fai-project.org.

CSEDU 2011 - 3rd International Conference on Computer Supported Education

520

and answering a few questions is not overwhelming

for small to moderate size labs, as it has to be done

only once.

Once a minimal installation is working on each

machine, everything else, from day-do-day mainte-

nance to installation of new operating system versions

or new software, can be done automatically, even for

heterogeneous hardware. A tool that we have found

invaluable for this purpose is Cfengine

. Usually the

lab runs for months with no or only minimal need of

manual intervention.

3.3 Beneﬁts

The LQE has led to a tremendous increase in produc-

tivity for students, and eliminated the amount of time

previously spent in installing software and transport-

ing ﬁles, with the following advantages:

1. Instant availability, local and remote, of data

which is not simple or practical to duplicate: mi-

crodata, digital cartography, etc.

2. Increased ﬂexibility in the ﬂow of information.

Students, from wherever they happen to be, can

spool ﬁles directly to their instructors’ desk, print

them in the lab’s dedicated printer or send them in

PDF format. This last option has been much used,

and affords entirely paper-free interaction. Com-

menting and/or grading of papers is accomplished

with PDF editors

3. Good interaction among students, while physi-

cally at the LQE.

It is true that interaction and cooperative learning

are not at the same level that the old-style computer

lab forced; we estimate that about 80% of the time

students work remotely, thus reducing their chances

of mutual assistance. We try to enhance student-to-

instructor and student-to-student interaction by ex-

ploiting forum facilities in e-learning platforms —like

Moodle

, one of the two adopted at our institution.

3.4 Validation

At the time of writing, the LQE has been in oper-

ation for over three years, and is therefore a well

established experience well beyond a proof of con-

cept. Similar experiences target energy efﬁciency,

(NComputing, Inc., 2010), enhancing active and co-

operative learning, (Moor, 2006), or web-based arran-

Both free and commercial versions exists; we have

used the free version, http://www.cfengine.org.

We use to this effect the free tool Jarnal,

http://jarnal.wikispaces.com; alternatives exist.

In http://moodle.org.

gements (see (Jalobeanu, 2006) and references

therein). Closer to our design goals is (Nixon

and Dwolatzky, 2002), although they mainly address

(back in 2002) presential labs.

4 CONCLUSIONS

Our experience shows that a dual-faceted lab, offering

both local and remote resources, is a feasible alterna-

tive, and relatively cheap to build with off-the-shelf

hardware and (mostly) free software. Its success in

our case has been critically dependent on a) Being

able to offer a user interface that looks almost exactly

the same, whether locally or remotely; b) The general

availability to students of wide band connections, and

c) A simple, “click-and-install” piece of software that

handles remote connections efﬁciently. The downside

is reduced student interaction, but use of fora (and, in

the future, chats), may go a long way to alleviate that.

ACKNOWLEDGEMENTS

Funding from grants IT-347-10 (Basque Government)

and ECO2008-05622/ECON (Ministry of Science

and Innovation, Spain) is gratefully acknowledged.

REFERENCES

Jalobeanu, M. (2006). Internet - a virtual laboratory for

distributed computing. In 2006 International Remote

Engineering Virtual Instrumentation Symposium.

Krafft, M. F. (2005). The Debian System (Concepts And

Techniques With Cdrom). No Starch Press.

Lamport, L. (1994). A Document Preparation System.

X User’s Guide and Reference Manual. Addison-

Wesley.

NComputing, Inc. (2010). Santa Ana College creates green

lab with virtual desktops.

R Development Core Team (2008). R: A Language and En-

vironment for Statistical Computing. R Foundation for

Statistical Computing. ISBN 3-900051-07-0.

Momjian, B. (2001). PostgreSQL : Introduction and Con-

cepts. Addison-Wesley.

Moor, S. S. (2006). Case study: Renovating a computer

teaching laboratory for active and cooperative learn-

ing. In Proceedings of the 2006 Illinois-Indiana and

North Central Joint Section Conference. American

Society for Engineering Education.

Neteler, M. and Mitasova, H. (2007). Open Source GIS: A

GRASS GIS Approach). Springer Verlag.

ENHANCING STATISTICS TEACHING WITH A VIRTUAL LAB - A Case Study of Seamless Local and Remote

Computing

521

Nixon, K. and Dwolatzky, B. (2002). Computer laboratory

infrastructure in engineering education-a case study at

wits. In Africon Conference in Africa, 2002. IEEE

AFRICON. 6th, volume 1, pages 431 – 436.

Stallman, R. M. (1997). GNU Emacs Manual. Free Soft-

ware Foundation.

CSEDU 2011 - 3rd International Conference on Computer Supported Education

522