A Cyber Laboratory for Device Dependent Hardware Experiments in

a Hybrid Cloud

Nobuhiko Koike

Faculty of Information and Computer Sciences, Hosei University, 3-7-2 Kajino-cho, Koganei-shi, 184-8584 Tokyo, Japan

Keywords: Remote Laboratory, Hardware Logic Design, Cloud Computing.

Abstract: The paper proposes a cyber laboratory in the form of a hybrid cloud, where the actual laboratory and the

remote laboratory are combined. At the existing physical experiment laboratory, the limitation of the

number of experimental devices for FPGA hardware design course resulted in platform usage congestions

and extended laboratory hours. Migration to the cloud should be the natural solution. However, existing

devices such as FPGA evaluation devices and Logic analyzers became obstacles to migrate to public clouds.

The proposed system combines an On-Premise private cloud organized by laboratory platforms to perform

device dependent services, and a public cloud where remaining design, development and evaluation stages,

are carried out in the form of a PaaS (Platform as a Service). The design and experiment tasks should be

modified accordingly to accommodate CAD tools in the set of the Web Services. Existing faculty database

server and educational support system combine the private cloud, the public cloud and the faculty servers, in

a seamless way. Students can migrate to and from laboratories at any design stages. As the device dependent

tasks have been implemented in the Web Services, efficient sharing of platforms can be achieved in space

and time sharing fashions. The public cloud ensures a scalable increase and decrease in server machines

according to the student usages and seasonal load changes. The laboratory managing software takes care of

the allocation and migration of student Virtual-Machines between the On-Premise private cloud and the

public cloud. It can also accommodate the BYOD (Bring Your Own Device) style student usage, by

preparing three different access methods: student side BYOD applications, the Web accesses and remote

desktop connections. The hybrid cloud approach achieves a scalable realization of the cyber laboratory

suitable for the BYOD style experiments, where efficient sharing of On-Premise laboratory platforms can be

realized in the mixed use of actual and remote laboratories.

1 INTRODUCTION

The author has been engaged in remote laboratory

projects and combining them to the actual

laboratories (Koike, 2012, Fujii and Koike 2008).

Hose University accepts around 100 new

undergraduate students each year to take the

hardware design course, utilizing twenty FPGA-

based design platforms including logic analyzers and

function generators. Students also share twenty PC-

based Logic analyzer/function generators. Allocated

time for the class is three hours per week. So, four

students share one platform as an average.

Schematic entry approach and Verilog-HDL design

approach are both employed in the coursework.

Each student is expected to design his/her own 32 bit

CPU as the final project. The 32bit CPU design by

making use of Verilog-HDL/logic synthesis tools,

implementation in a FPGA evaluation board and

verification of the designed hardware by running test

software, are complex tasks and time consuming but

it is effective for undergraduate students to acquire

enough knowledge and skills in the field. Such a real

life experience in the actual laboratory is very

important and rewarding for developing student

carrier. It is desirable for students to continue their

experiments outside the actual laboratory. The

author started the remote laboratory project, and

tried to combine the remote laboratory with the

actual laboratory. The former effort to access remote

experiment sites (Hassan, 2012) only achieves the

functionality of remote experimental data access via

Web browsers. It had no control of remote

experiment devices. Also, there are many efforts to

migrate campus environments to the cloud

computing, e.g. (Kumar, 2012; Hassan, 2012). In

such a cloud configuration, the remote desktop

services should be employed for remote experiment

440

Koike N..

A Cyber Laboratory for Device Dependent Hardware Experiments in a Hybrid Cloud.

DOI: 10.5220/0004942504400445

In Proceedings of the 6th International Conference on Computer Supported Education (CSEDU-2014), pages 440-445

ISBN: 978-989-758-020-8

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

device accesses. However, they have their scalability

problems due to the exclusive use of On-Premise

laboratory devices. The author’s former

implementation (Koike, 2013) adopted the On-

Premise private cloud solution. However, the

laboratory platform/servers become a bottleneck, if

the number of simultaneous remote laboratory users

is increased. It is also difficult to support recent

BYOD style student usages, as the former system

heavily relied on student side powerful laptops,

where time-consuming, latency-sensitive tasks can

be executed locally by offloading them from the

laboratory platforms. However, BYOD devices

could not meet such performance requirements.

Addressing these shortcomings, the author started

the new cyber laboratory project, which combines

the On-Premise private cloud and a public cloud in a

seamless way, namely the Hybrid Cloud solution.

On-Premise private cloud computers only perform

device depending services in the form of the Web

Services. The remaining services have been

migrated to public cloud computers. If a student is

accessing through the university-leased high-end

laptop, it can still exploit student’s laptop PC powers,

by migrating time-consuming services, such as logic

entry, simulation and verification to the laptop, in

the form of native Windows applications. On the

other hand, for the BYOD case, the allocated Virtual

Machine in the public cloud acts as a BYOD proxy

for the user. The proxy performs most of the works

and the communications with the BYOD can be

carried out via remote desktop, BYOD applications

or http connections. In this way, it can achieve a

scalable increase in the number of public cloud

computers, according to the students’ actual usages,

and also it can decrease during off-seasons. It can

drastically reduce the TCO (Total Cost of

Ownership).

By combining those technologies, highly

scalable and flexible cyber laboratory can be

realized.

2 SYSTEM CONSIDERATIONS

FOR CYBER LABORATORY IN

A HYBRID CLOUD

In order to give students real life laboratory

experiences, it is important to realize almost the

same laboratory environment both in remote and

actual laboratory modes. The efficient sharing of the

laboratory platforms and their devices becomes

important for realizing a scalable Cyber Laboratory.

In case of the remote laboratory mode, most

experiment tasks can be offloaded on the public

cloud, except device related works, such as FPGA

setup/run or logic-analyzer setup/get results.

Thanks to the advancement in cloud computing,

implementing the laboratory computer environments

in a public cloud becomes advantageous to reduce

total cost of ownership. However as said, laboratory

devices, such as Verilog-HDL synthesis tools,

FPGA evaluation platforms, Logic Analyzers and

pattern generators have made it difficult to migrate

to the public cloud. Instead, the author’s previous

system, have chosen an On-Premise private cloud

solution. On-Premise private cloud allows any

connections of propriety devices to the On-Premise

laboratory platform computers to organize as a

private cloud. Although it achieved an efficient

sharing of laboratory platforms/servers, it has its

own scalability problem. If the number of remote

laboratory users is increased, the server becomes

overloaded and resulted in higher latencies and

longer elapsed time. The former system also took

advantage of students’ high-end laptop PC

performances, to offload time consuming tasks to

the laptops. Thus, effective offload of laboratory

platforms was realized. However, this approach

becomes difficult to adopt for recent trend of BYOD

(Bring Your Own Device) style student usages.

Usually, BYODs are rather poor in CPU

performance. It is better to offload such laboratory

PC loads to computers in a public cloud. So, a

hybrid cloud organization, where the On-Premise

cloud performs device dependent tasks and the

public cloud performs rest of the works, allows the

realization of a flexible and scalable Cyber

Laboratory.

The Figure 1 shows the Cyber Laboratory

System organization, which have been designed

based on the followings design considerations:

-Use of Special devices in the On-Premise private

cloud: The requirement to connect specialized

devices became an obstacle to choose a public cloud

solution. Instead, an On-Premise private cloud

solution has to be employed. However, the amount

of workloads should be as little as possible. Only,

device-related tasks such as FPGA Load/RUN,

Logic-analyzer setup /measurement /results, should

be remained in the laboratory platform (private

cloud computer). The rest of the workloads should

be offloaded on the public cloud. Only when device

related services have been requested, the Web

services should be generated and sent to the

laboratory platforms. In order that such device

dependent tasks can become accessible through the

ACyberLaboratoryforDeviceDependentHardwareExperimentsinaHybridCloud

441

internet, they should be first transformed into

Windows applications, and then encapsulated in the

form of the Web services. If a student virtual

machine is resided in the same laboratory platform,

it realises an actual laboratory environment. The

virtual network connection provides enough

bandwidth to connect them. Nowadays, various

system software for building such a private cloud,

are available (Hasan 2012). So, the system

development efforts and system management works

can be remarkably reduced. The system manager

only set the policies and the rest of the works can be

handled semi-automatically. Students can find out an

appropriate platform and log-in through the “dash-

board” monitoring page of the Cyber Laboratory

Manager in a self-serviced manner. It performs VM

allocations/ migrations, server assignments and

cloud storage link setup/ synchronization.

-Use of public cloud for handling non-device-related

services: Non-device-related tasks can easily be

offloaded to the public cloud, once they have been

transformed into Windows applications and again

encapsulated in the Web services. For each student

one virtual machine is allocated in the public cloud,

thus the remote laboratory environment is realized.

In case when device-related services have been

requested, these services should be forwarded to one

of the private cloud computers in the remote

laboratory service mode.

The use of the public cloud allows an efficient

computer usage, as the number of actual computers

can be flexible, according to the student usages and

seasonal load variations. Each student is allocated an

individual virtual machine, which can be migrated

between private and public clouds, according to the

actual /remote laboratories or mixture configurations

of them.

-Adoption of the BYD style student use: The cyber

laboratory managing software should detect the

student BYOD devices and set up an appropriate

service either in the private or public clouds. If the

student is using the University-leased laptop, most

design tasks, except the license logic synthesis tool

and actual FPGA device use, can be offloaded on the

laptop, and the student can enjoy a comfortable

experiment. In case when Logic synthesis and FPGA

run services have been requested, the resident

software issues the corresponding Web services to

Figure 1: Cyber Laboratory in a Hybrid Cloud Organization.

CSEDU2014-6thInternationalConferenceonComputerSupportedEducation

442

the On-Premise private cloud. In case when the use

of non-laptop BYOD is specified, one virtual

machine is allocated as a proxy in the public cloud.

The proxy performs most laboratory works. Only

when the device related services have requested, the

request is forwarded to one of the platform servers in

the private cloud, via the Web services.

-Cyber Laboratory managing software handles

allocations of the actual and remote laboratory

Virtual Machines (VMs): It manages all setups of

laboratory machines (private cloud), and the remote

laboratory machines in the public cloud. The Cyber

Laboratory manager serves as the portal of the

Cyber Laboratory. After usual login, and VPN

setups, it handles each student’s VM allocation,

which can be done through “dashboard”. As it shows

current actual and remote laboratory work-load

status, students can easily select either actual or

remote laboratories. Also, the user can ask the

manager for the VM migration to/from

actual/remote laboratories. Once the VM setup

completed, the manager redirects the request to the

designated VM/server. If the student’s BYD is

specified, the allocated VM performs not only usual

experiment works but also performs the proxy

services for the BYOD. The customized remote

desktop/application or Web browser in the BYD will

communicate with the proxy server in the VM. The

course manager just set laboratory platform usage

polices, according to the office-hours/ off-hours. The

laboratory platform allocations can easily be realized

by making use of available VM allocation software.

The ratio of the actual/remote laboratory platforms

can be adjusted according to the platform usage.

-CAD design file sharing among Private/Public

cloud, faculty database and faculty educational

system: In order to realize efficient sharing and

transferring huge CAD data, the faculty database

(file server) serves as a kind of cloud storage. As all

the Web services send/receive only file IDs, an

efficient data transferring can be achieved.

3 CYBER LABORATORY IN A

HYBRID CLOUD

ORGANIZATION

The proposed Cyber laboratory consists of On-

Premise Private Cloud and the public cloud.

Between two clouds, the Web services are employed

as the communication means. It can combine the

actual laboratory and the remote laboratory, with

newly designed CAD services and FPGA-run

services realized in the form of the Web Services.

The faculty database plays an important role in Data

sharing and synchronizations. It enables to share

large experiment data among laboratory platforms,

student/Teacher/TA PCs. As large CAD design files

can become accessible from any PCs, the web

services only need to transfer a few CAD parameters

and the design file location IDs. Thus the file

transfer overhead can be minimized. It is also useful

for students, TAs and Teachers for further

consultations and evaluations. Students can easily

migrate from actual laboratory to remote laboratory

and vice versa. And students can continue their

projects at home, using their own PCs in almost the

same environment as in the actual laboratory

environment. Figure 2 show the private cloud

system organization. For each laboratory experiment

platform, which is one of the On-Premise private

cloud components, a Web service virtual machine is

allocated. It contains integrated FPGA tools, license

Verilog-HDL synthesizer and the Web service

server. The integrated FPGA tools accept the FPGA

related Web services and perform device related

tasks, such as FPGA Set-up/Run, logic-analyzer

setup/run/get results. The Verilog-HDL logic

synthesizer is the license software, encapsulated in

the Web services. It accepts high-level descriptions

and returns synthesized Verilog-HDL descriptions.

Figure 2: On-Premise Private Cloud System Organization.

The synthesizer is not a “must”. Although

students can do experiments without them, it can

reduce significant amount of design works by

combining high-level design and use of such logic

synthesis tools. The synthesizer is not affordable to

install all students’ laptop or public cloud computers.

So, the license software are only installed in the On-

Premise private cloud laboratory platforms and the

limited number of the public cloud computers. By

preparing the Web services for the logic synthesis,

students who want to use this functionality, can use

them from remote site.

ACyberLaboratoryforDeviceDependentHardwareExperimentsinaHybridCloud

443

In the case of actual laboratory mode, student virtual

machines are also set up on the corresponding

laboratory platforms. The virtual network switch

allows rapid inter virtual machine connections, as

they reside on the same computer and messages do

not go out of the computer. In this way, plural

students’ experiments can be carried out

concurrently, by sharing FPGA and logic analyzer

devices, even in the actual laboratory.

The cyber laboratory manager performs the

system management tasks, such as student

platform allocations in On-Premise private cloud

and the public cloud setups. The course manager set

the policy to the cyber laboratory manager, and the

rest of works can be done in self serviced fashions.

The dash board service lets students to know current

workload status and to select available platforms.

Figure 3 shows the public cloud side virtual

machine organization. After the public cloud setup is

completed, physical servers become available. The

limited number of them should be assigned to run

the logic synthesis virtual machines. Remaining

servers are pooled for the use of student virtual

machines. The purpose of the logic synthesis VMs is

to offload the logic synthesis works from the

laboratory platforms and to run the services in the

public cloud. As the license limits the number of

VMs, student VMs have to be installed the

integrated FPGA tools without license Verilog-HDL

synthesis tool and FPGA setup/RUN tools. Those

removed tools are replaced with the corresponding

Web services.

The logic synthesis VM organization is rather

simple. It accepts the logic synthesis Web service

requests, these are generated from other student

VMs. In order to get/set the Verilog-HDL

descriptions, the FTP is setup with the faculty data

base (Experiment Data server).

Each Student VM handles the remaining

laboratory works, except the device related services

and the license Verilog-HDL synthesis. The student

VMs generate corresponding Web services and send

them to the designated VMs: Logic synthesis VMs

in the same public cloud or the Web service VMs in

the On-Premise Cloud via the internet. Another

student VM’s role is to act as the proxy for the

student. If the student specifies the use of BYOD,

then BYOD Proxy establishes the connection with

the BYOD, via the remote desktop, BYOD

application or Http.

The Figure 4 shows the experiment flows in

actual and remote laboratories. For each student, a

student virtual machine is allocated. By allocating

Figure 3: Public Cloud Side Virtual Machine Organization.

the student VM in the same laboratory platform as

the Web Service VM, the actual laboratory

environment is realized. All design stage works are

carried out within the same platform.

In case of the remote laboratory mode, the

student VM is allocated in one of servers in the

public cloud. If the FPGA run is requested, the

student VM setups the connection with one of the

FPGA Web service VM in the On-Premise private

cloud to do the FPGA run.

Figure 4: Experiment Flows in remote and actual

laboratories.

In case of the logic synthesis request is issued, the

CSEDU2014-6thInternationalConferenceonComputerSupportedEducation

444

student VM setups with one of the logic synthesis

VMs in the public cloud. The corresponding logic

synthesis Web services connect VMs via the public

cloud intra-net. In this way, almost the same

experiment environment can be realized both in

actual and remote laboratory modes.

4 CONCLUSIONS

The Cyber laboratory in a hybrid cloud project

combines the actual laboratory and the remote

laboratory in a seamless way. It gives remote users

real life experiences in hardware logic design and

implementations. The device related services are

handled by one of laboratory platforms as a private

cloud.

The remaining non-device-related tasks are

handled by student virtual machines in the public

cloud. It can still exploit student side PC powers to

offload platform workloads, if high-end PCs are

available. In case when a student should login

through a BOYD, a virtual machine in the public

cloud is allocated and it serves as a proxy. The use

of available cloud computing system tools

contributed to reduce system developments works as

well as system management/operation works.

The use of the experiment server as a Web

storage contributed to realize a seamless migration

between remote and actual laboratories. It also made

it possible to share experiment data among students,

TAs and teachers for further consultations. So, a

flexible and scalable Cyber Laboratory, which can

reduce the TCO, can be realized.

The project is just started and it is rather early to

evaluate the effectiveness of the proposed system,

but the author believes its success. The remaining

problem is how to decouple the hardware integrated

software, and to combine public cloud computing

environments.

REFERENCES

Koike, N., 2013. A Cyber Laboratory as a Private Cloud:

Combining Actual Laboratory and Remote Laboratory

for Hardware Logic Experiment. In International

Conference on Engineering Education and Research

(iCEER2013).

Koike, N., 2012. Cyber Laboratory for Hardware Logic

Experiments: A Seamless Integration of Actual

Laboratory and Remote Laboratory. In 2012

International Conference on Information Technology

Based Higher Education and Training (ITHET 2012).

Fujii, N., Koike, N., 2008. Work in Progress - A Dual

Mode Remote Laboratory System Supporting Both

Real-Time and Batch Controls by Making Use of

Virtual Machines. In 38

ASEE/IEEE Frontiers in

Education Conference.

Kumar, N., Mittal, R.K.,2012. Cloud Computing Setup for

a Campus Environment. In proc. International

conference on Cloud Computing, Technologies,

Applications & Management.

Hassan, S. A. G., 2012. STAR: A Proposed Architecture

for Cloud Computing Applications. In proc.

International conference on Cloud Computing,

Technologies, Applications & Management.

Bauer, M., et al, 2012. Experimenter’s Portal: Collection,

Management and Analysis of Scientific Data from

Remote Sites. In proc. MGC2012.

ACyberLaboratoryforDeviceDependentHardwareExperimentsinaHybridCloud

445