Distance Learning and the Internet Conference 2008

5-2
Distance Learning Computer-based Hands-on Workshop: Experiences on Virtual and Physical Lab Environments
Patcharee Basu, Shoko Mikawa, Achmad Basuki, Achmad Husni Thamrin, Keiko Okawa, Jun Murai Keio University

Plenary Sessions
Waseda University Presentations

{yoo, funya, abazh, husni, keiko, jun}@sfc.wide.ad.jp

Abstract
In response to the educational demands of computer-skilled human resources, distance learning with ability to support hands-on computer lesson is needed. In this paper, a platform for region-wide distance learning computer-based hands-on workshop is presented through the actual developments. The proposed platform supports 1) teaching/learning activities in a hands-on computer workshop 2) efficient large-scale remote computer laboratory. Computer virtualization and StarBED large-scale computing testbed were utilized to create a distance learning computer laboratory, virtual and physical. This paper discusses various aspects of deploying virtual and physical lab environments for region-wide learners in a synchronous-style distance learning workshop.

Keywords
Hands-on computer workshop, remote laboratory, distance learning, educational technology

1 Introduction
With digital and telecommunication technologies, distance education has been developed to extend boundary of knowledge sharing to be more location and time independent. It has been widely deployed in academic or training programs to mobilize knowledge within or across organizations, countries and regions. With its characteristic to share knowledge to more learners at remote locations, it has become a costeffective solution for human resource development. In the digital age which daily operations of humans are computer-assisted, practical skills to use computers to accomplish specialized tasks in different fields are required and included in recent education. Nonetheless, most distance education systems still lack the ability to provide hands-on computer lessons. Therefore, researches and studies to combine computer laboratory for hands-on study in distance education are necessary to enhance learning opportunities beyond its lecturing ability.

In this paper, we present a study on combining remote computer laboratory in a distance education system through an actual experience of School on Internet Asia (SOI Asia)(Mikawa,2006). SOI Asia is a consortium of 27 universities and research institutes in 13 Asian countries as of September 2008. The project started in 2001 to develop a distance education platform among Asian universities and has conducted more than 30 graduate level courses(289 lectures) to region-wide students. For autonomous operation of distributed learning sites on the platform, staffs at all sites need to be trained to operate their local facilities to run distance education program. Therefore, SOI Asia has held annual operator workshops from 2002, as shown in Table 1, for intensive hands-on training of region-wide participants. The workshop program focuses on topics related SOI Asia network and application technologies as shown in Table 2. Some details of hands-on lessons, e.g. operating system(OS) versions, software and network topologies, could be slightly different each year depending on workshop instructors and recent technologies on the SOI Asia platform. The workshop model changed from face-to-face model to distance learning model in 2006 concerning the scalability as discussed in Section 3. The challenge to develop a distance learning computer-based hands-on workshop is to design 1) a platform to facilitate teaching/learning activities in a hands-on computer workshop which differ from lecture and conference 2) a computer laboratory which can remotely and efficiently support the roles of class members, including instructors, teaching assistants(TAs) and participants. Their roles need to test, deploy, administer, execute, monitor and troubleshoot on the laboratory. These design considerations were addressed in the implementations of SOI Asia distance learning workshops in 2006 and 2008. The system to facilitate teaching/learning workshop is presented in Section 4.1. The computer virtualization and StarBED (Miyachi, 2006) technologies applied to create remote
111

DLI Review & Report

Session 1-3 Session 4-6 Session 7-9

Session 5 Global Educational Networking/e-Pedagogy

computer laboratory is discussed in Section 4.2. This paper emphasizes on sharing actual experiences gained from deploying virtual and physical laboratory. The characteristics including advantages and limitations of both technologies will be discussed in Section 4.2. The ordinary laboratory deployment in the face-to-face model of SOI Asia workshop will be discussed to provide background information in Section 3. Table 1: SOI Asia operator workshops in the past
Year Model Location 2002 2003 2004 2005 2006 2008 F2F F2F F2F F2F DS DS 1 1 2 1 15 16 Country Japan Thailand Indonesia Thailand Indonesia 10 Asian countries 10 Asian countries Day Participant 5.5 7 6 6 5 5 5 21 21 13 14 33 42 42

support remote hands-on computer environment. Remote console access technologies including VNC and Windows remote desktop were deployed (Summers,2008). Computer virtualization technology (Nieh,2006.,Rigby,2006) were utilized for hands-on environment. Further experiments and researches in this area on different technologies will be beneficial for future developments.

Plenary Sessions
Waseda University Presentations

3 Face-to-Face workshop model
The basic model of hands-on computer workshop is the face-to-face model which class members and hands-on equipments are at the same location. During 2002-2005, each year SOI Asia organized a face-toface workshop by inviting instructors and participants to travel to a host institution in Asia for 5-7 days as shown in Table 1. Instructors have prepared and tested lab lessons earlier at their computer laboratories and brought necessary files for laboratory deployment, e.g. OS image, software and configuration files. Equipments, proportional to the number of participants, including computers and network switches were temporarily rent at the workshop site in most cases. Instructors and TAs prepared those equipments 1-2 days before the workshop started by manually installing and configuring necessary hardware, OS images and applications. Since it was difficult to afford the exact specifications for all equipments, there were cases of hardware incompatibilities with instructors’ settings. To create a desired network topology, computers were manually interconnected through network switches. According to lab lessons, each day the configurations of computers and networks might need to be changed. The face-to-face workshop well achieved its objective to train skilled human resources. One strong advantage is that participants can study physical administration of devices and can learn to setup system from scratch, this physical experiences could hardly be obtained in the distance learning model. However, the face-to-face model could not scale to support the increasing number of workshop participants due to the increasing costs of travels and the increasing workload to manually manage computer laboratory(Mikawa,2007). With larger number of physical equipments, manual administration was inefficient and time-consuming. There were more possibilities of hardware failure and hardware incompatibility problems. Therefore, alternative models to implement a computer-based hands-on workshop with better scalability and efficiency should be explored for future expansion.

DLI Review & Report

Note: F2F is Face-to-Face model DS is Distance-Learning model

Table 2: Topics taught in the workshop
Topic OS Number of network topology

Session 1-3 Session 4-6 Session 7-9 112

1. TCP/IP basic and Unicast routing 2. Multicast routing 3. SOI router installation and configuration 4. SOI server installation and configuration 5. SOI classroom applications

FreeBSD 1-2 FreeBSD 1 FreeBSD 1 Linux Windows 1 1

Note: 2002-2005 – IPv4 based technologies, from 2006 – Dual-stack and IPv6 based technologies.

2 Related works
There are several distance learning development programs that focus on sharing educational content region-wide or world-wide, for example, Global Development Learning Network(GDLN), Association of Pacific Rim Universities(APRU) (Allen,2003), ASEAN Virtual Institute of Science and Technology (AVIST)(Kanchanasut,2004). The established E-learning environments are designed mainly for lecturing and conferencing activities, therefore they are not equipped with a system to support computerbased hands-on trainings. On the other hand, several works were proposed to

Distance Learning and the Internet Conference 2008

Plenary Sessions
Waseda University Presentations

Figure 1: Instructors and TAs were preparing equipments in a face-to-face workshop

4 Distance learning workshop model
The distance learning workshop model deployed in 2006 and 2008 is composed of two main components. One is a system to facilitate teaching/learning activities in a hands-on computer workshop from class members at distributed locations in Asia. The existing SOI Asia distance education platform was utilized with some newly developed tools that support additional activities in a hands-on computer class, as presented in Section 4.1. Another is the computer laboratory which can remotely and efficiently support different tasks of class members which details will be discussed in section 4.2.

Figure 2: SOI Asia network platform for distance learning workshop Applications have been designed to support the three main activities in a computer-based hands-on workshop; conversation, presentation and demonstration. Conversation among class members is provided in the many-to-many model by video, audio, textbased chat as shown in Figure 3. The text-based chat system is suitable for precisely exchanging text content in hands-on lessons, e.g. Unix commands and configurations. Presentation is carried out by a webbased flash mechanism which workshop instructors share synchronized views of slides with animation and handwriting to participants. Demonstration to illustrate some working examples or experiments by instructors is considered compulsory in this computer-based hands-on workshop. SOI Asia developed a tool to lively display instructors’ Unix terminal outputs to web viewers, as shown in Figure 3. It has simpler access and clearer view compared to transmitting instructor’s computer screen as a video stream experienced in the workshop 2006.
DLI Review & Report

Session 1-3

4.1 Distance learning platform
The SOI Asia distance education platform was designed to be practical and to achieve quality and scalability of both network and application. Instead of totally relying on the existing Internet with inadequate quality to deliver distance education traffics (Basu,2007), the hybrid network of a dedicated unidirectional broadcast satellite link and the existing Internet at local universities are used as shown in Figure 2. Lecturer site can be at any locations provided that the Internet to the satellite hub station, called the gateway site, can accommodate high-quality lecture traffics. Lecture traffics are converted to the acceptable quality and formats before being relayed onto the satellite link to reach all student sites in Asia. At student sites, they can transmit feedbacks to lecturer and other student sites using their local Internet connectivity. The feedbacks can be video, audio, text chat and webboard post depending on the available bandwidth at local site. Unidirectional link routing(UDLR)(Duros,2001) and IPv6 Multicast technologies were deployed to provide integrated network environment with optimized bandwidth consumption and scalability.

Session 4-6 Session 7-9

Figure 3: Applications in SOI Asia workshop
113

Session 5 Global Educational Networking/e-Pedagogy

4.2 Remote computer laboratory
The design of remote computer laboratory focuses on both its functionality and efficiency by considering the requirements as shown in Table 3. In terms of functionality, it must support instructors’ lab designs including specifications of computers and network topologies. The ability to connect lab network to Internet is desirable to introduce a more realistic lessons. Class members must be able to remotely perform their roles as stated in Table 3. The response times of these remote accesses are crucial because it directly affects participants’ hands-on experiences and the whole class progress. The efficiency of lab administration to manage large-scale laboratory with dynamic changes of configurations must be considered. Failure to keep lab integrity according to instructors’ lab designs will disturb the class. Considering these aspects, SOI Asia has designed and deployed both virtual and physical schemes of remote computer laboratory. Virtual laboratory by the computer virtualization technology is discussed in Section 4.2.1. Physical laboratory by StarBED computing testbed is discussed in Section 4.2.2. Table 3: Requirements of remote computer laboratory
Lab design - Number of equipments - Hardware/software specifications - Network topologies - Connectivity to Internet

functions like a physical machine, e.g. runs an OS and communicates with other hosts. Mechanisms to manipulate these VMs through remote connection are commonly provided by virtualization products. Figure 4 illustrates the deployment of computer virtualization in the SOI Asia workshop 2006. Seven virtualization servers were setup to create a virtual laboratory for 42 participants at 15 locations in 10 Asian countries. Each server hosted 6-12 VMs depending on their performance. Each participant was assigned a VM and used a local PC to connect to his/her VM. They can see and manipulate VMs by remote console.

Plenary Sessions
Waseda University Presentations

DLI Review & Report

Figure 4: Deployment of computer virtualization The assessment of deploying computer virtualization technology for computer-based hands-on workshop is discussed based on SOI Asia experiences using VMware product as shown in Table 4. The instructors’ lab designs should be actually tested in order to use a computer virtualization product, although they are supposed to support general lab design with common computers and network switches. SOI Asia lab designs were tested with Xen product in 2006 but it was not stable to run a lab using unmodified FreeBSD. Therefore, VMware product was tested and used to implement the virtual laboratory. VMware has a server console application which provides a mechanism to access and monitor console of VMs through remote connection. With concern of response time, the bandwidth consumption of this application was measured too high and could yield slow response time in our environment considering the number of concurrent participants’ connections and the satellite link capacity. As a result, another mechanism to forward console traffics through remote connection with lower bandwidth consumption was developed for participants’ console accesses(Basuki, 2007). Other factors affecting the response time are the overloaded virtualization servers and the underlying Internet quality(Basuki, 2007). Since resources on a virtualization server are shared by a number of VMs, during high-load hands-on activities, e.g. each participant run a high-computing process, the CPU utilization of the virtualization

Session 1-3 Session 4-6 Session 7-9 114

Remote access for - Lab test and deployment class members’ roles by instructors/TAs - Hands-on works by remote participants - Monitor and troubleshoot participants’ lab works by instructors and TAs - Remote administration from TAs - Good response time Administration - Efficiency to deploy lab scenarios and maintain large-scale laboratory - Accuracy to provide lab integrity according to the given lab designs

4.2.1 Virtual laboratory
Computer virtualization technology was applied to create hands-on computer laboratory for SOI Asia workshop year 2006(Basu,2007). A physical machine which runs virtualization service, called virtualization server, can create a number of virtual machines(VMs) with the required virtual hardware specifications, e.g. CPU, RAM, Harddisk and also can create virtual network to connect these VMs. Each VM can perform

Distance Learning and the Internet Conference 2008

server could be very high causing slower response time. Therefore, this critical condition should be taken into account to determine the appropriate number of VMs to be hosted on a virtualization server. In addition, congested Internet link causes slow response time, as it was the case in 2006 which the Internet links from some partner sites were congested. In the aspect of laboratory administration, provided with automated and soft-configuration means to handle VMs, the lab deployment and maintenance is fast and accurate. Reduced number of physical machines for the same laboratory size saves costs to afford computers and reduces human workload. Accordingly, efficiency to manage large-scale laboratory can be achieved in the virtual laboratory. Table 4: Characteristics of deploying computer virtualization as a remote hands-on computer laboratory
Lab design - General lab designs are supported but a test is needed to avoid any incompatibilities - Supports lab connectivity to Internet - Lower performance lab computers

pants at 16 locations in 10 Asian countries. A cluster of 168 StarBED nodes was allocated with specifications according to Table 5. The network interfaces of these StarBED nodes are connected to a network switch cluster which can be programmed to create different lab topologies and connect to Internet. This experiment network is separated from the management network to avoid interference between experiment and access/control traffics. Remote access from class members are made through a gateway and then StarBED nodes by SSH (Ylonen,2006). Table 5: Allocated StarBED nodes’ specifications
Model Chipset, CPU Memory, Disk Network Interfaces NEC Express5800/110Rg-1 Intel E7230, Pentium4 3.2GHz(HT) 8 GB, SATA 80GB x 2 INTEL Pro/1000 PM Server x 2 INTEL Pro/1000 MT Dual Port Server x 2

Plenary Sessions
Waseda University Presentations

DLI Review & Report

Remote access for - The required class members’ class members’ accesses are supported roles - Response time can be slow due to overloaded server, congested Internet, highbandwidth remote connection Lab administration - High efficiency and accuracy to deploy lab scenarios and maintain large-scale laboratory - Reduced number of computers - Less physical problems, e.g. hardware incompatibilities and failures

Session 1-3

Figure 5: Deployment of StarBED environment The assessment of deploying StarBED for computer-based hands-on workshop is discussed based on SOI Asia experiences as shown in Table 6. In terms of functionality, the high-performance machines in StarBED can better support lab designs which require high computing power. With extra network interfaces, they can support more functions than only being a host, e.g. router. Another advantage of being physical environment is that if a lab lesson needs any special devices to be attached to computers, it is more straightforward to apply. However instructors need to design labs within the limits of hardware specifications, while it could be more flexible in computer virtualization case. Thorough test of lab designs on the StarBED environment is required to make sure there is no hardware incompatibility problem, however the identical hardware set eases the verification task. The remote access to StarBED nodes depends on the system deployed on the nodes, e.g. using SSH to Unix
115

Session 4-6

4.2.2 Physical laboratory
StarBED environment(Miyachi, 2006) was applied to create hands-on computer laboratory for SOI Asia workshop year 2008(Basu,2008., Miyachi, 2008). StarBED is a large-scale computing testbed aiming to provide realistic computer and network environment for research and educational experiments. The testbed is consisted of 680 high-performance computers and network switch clusters. It is located at Ishikawa prefecture, Japan and operated by National institute of information and communications technology(NICT). Figure 5 illustrates the deployment of StarBED in the SOI Asia workshop 2008. A physical computer laboratory was created to support training of 42 partici-

Session 7-9

Session 5 Global Educational Networking/e-Pedagogy

lab computers. The response time of remote connection is improved compared to the computer virtualization case because a computer’s resources are dedicated to single participant. However, the quality of underlying Internet still can be a factor that affects the response time. For the remote administration of StarBED nodes, lab deployment is supported by several automated processes which are fast and accurate(Basu, 2008) compared to manual administration of large-scale physical laboratory. The possibility of hardware fault or incompatibility is higher than virtualization case. For the administration tasks, StarBED provides some mechanism for remote controls including power management and console access(Miyachi,2008). However, some tasks which really require physical accesses, e.g. replaces hardware, could not be done by any class members, but local StarBED staff. Some backup nodes are prepared to handle the case. Therefore, the remote administration is more controllable in virtualization case. Table 6: Characteristics of deploying StarBED environment as a remote hands-on computer laboratory
Lab design - Lab designs need to consider the specifications of StarBED nodes - Possibility to add special devices - High performance lab nodes - A test is needed to avoid hardware incompatibilities

nities, especially to regions with inadequate educational resources. The proposed platform supports teaching and learning activities of hands-on computer workshop by applications supporting conversation, presentation and, particularly for workshop, demonstration. The underlying network is designed to support regionwide learners with quality and scalability. For the remote computer laboratory, both computer virtualization and StarBED have different characteristics in the deployment. Their ability to support remote accesses from class members enhances distance learning to include hands-on computer lessons in addition to theoretical study. They have better efficiency and accuracy to manage large-scale laboratory compared to the ordinary manual administration. The physical laboratory provided by StarBED has its advantages of high and dedicated computing resources but using actual hardware needs to be careful with hardware incompatibility and fault problems. The virtual laboratory provides flexible lab designs to instructors and full control to lab administrator because VMs are softconfigurable. There are less hardware concerns. The load on each virtualization server should be considered in the design to balance between utilization and user experiences. These trade-offs are useful to be considered in the future workshop designs to decide a suitable solution for other workshop contexts.

Plenary Sessions
Waseda University Presentations

DLI Review & Report

References
Allen, K. & Poh-Kam, W. (2003). Technology spinOffs from pacific rim universities: entrepreneurial context and economic impact. Proceedings of the 23rd Annual Entrepreneurship Research Conference. Basu, P., Mikawa, S., Basuki, A., Thamrin, A.H. & Okawa, K. (2008). An Asia-wide distributed hands-on workshop: synchronous learning and large-scale computing laboratory. Proceedings of Computers and Advanced Technology in Education. Basu, P., Mikawa, S., Basuki, A., Thamrin, A.H., Okawa, K. & Murai, J. (2007). Combination of online virtual computer laboratory and region-wide distance learning for IT educat i o n i n A s i a . P ro c e e d i n g s o f Wo r l d Conference on Educational Multimedia, Hypermedia and Telecommunications, 1261-1269. Basu, P., Mikawa, S., Thamrin, A.H., Okawa, K. & Murai, J. (2007). Internet technologies and infrastructure for Asia-wide distance education. Proceedings of International Symposium on Applications and the Internet, 3-9. Basuki, A., Thamrin, A.H., Okawa, K. & Murai, J.

Session 1-3 Session 4-6 Session 7-9 116

Remote access - Remote access mechanism depends on the system for class deployed on StarBED nodes members’ roles - Performance does not affect slow response time but still depends on congested Internet, if any. Lab administration - High efficiency and accuracy to deploy lab scenarios - Better efficiency to administer large-scale physical machines compared to manual laboratory administration - Remote administration is less controllable than computer virtualization - Hardware fault possibilities is higher than computer virtualization

5 Conclusion
With increasing demands of computer-skilled human resources, recent education includes hands-on computer lessons for learners to obtain both practical and theoretical proficiency. Distance learning with extended capability to support computer-based hands-on lesson can enhance more learning opportu-

Distance Learning and the Internet Conference 2008

(2007). A remote hands-on exercise environment for an Asia-wide real-time workshop. Proceedings of International Symposium on Applications and the Internet Workshops, 38-40. Duros, E., Dabbous, W., Izumiyama, H., Fjuii, N. & Zhang, Y. (2001). A link layer tunneling mechanism for unidirectional links, RFC 3077. http://www.ietf.org/rfc/rfc3077.txt. Kanchanasut, K., Nualchawee, A. & Arunwatanamonkol, P. (2004). Science and technology human resource development:E-learning initiative for ASEAN. Proceedings of eLearningAP 2004, 209-212. Mikawa, S., Basu, P., Tsuchimoto, Y., Okawa, K. & Murai, J. (2006). Multilateral distance lecture environment on the Internet for Asian universities. The journal of Information and Systems in Education. 5,84-93. Mikawa, S., Basu, P., Okawa, K. & Murai, J. (2006). An Asia-wide realtime distributed hands-on workshop. Proceedings of International Symposium on Applications and the Internet Workshops (SAINTW’07), 41. Miyachi, T., Chinen, K. & Shinoda, Y. (2006). StarBED and SpringOS: large-scale general purpose network testbed and supporting softw a r e . P ro c e e d i n g s o f I n t e r n a t i o n a l Conference on Performance Evaluation Methodologies and Tools. Miyachi, T., Basuki, A., Mikawa, S., Miwa, S., Chinen, K. & Shinoda, Y. (2008). Educational environment on StarBED. Proceedings of Asian Internet Engineering Conference. Neih, J. & Vaill, C. (2006). Experiences teaching operating systems using virtual platforms and Linux. ACM SIGOPS Operating Systems Review. 40(2), 100-104. Rigby, S. & Dark, M. (2006). Designing a flexible, multipurpose remote lab for the IT curriculum, SIGITE ‘06, Proceedings of the 7th conference on Information technology education. 161-164 Summers, W.C., Bhagyavati & Martin, C.(2008). Using a virtual lab to teach an online information assurance program, Proceedings of the 2nd annual conference on Information security curriculum development. 84-87 Ylonen, T. & Lonvick, C. (2006). The secure shell (SSH) authentication protocol, RFC 4252. http://www.ietf.org/rfc/rfc4252.txt.

Plenary Sessions
Waseda University Presentations

DLI Review & Report

Session 1-3 Session 4-6 Session 7-9 117