A Framework for Resource-Sharing Virtual Laboratory Architecture

The Nigeria Computer Society (NCS), 3rd Research Consortium on InformationTechnology Innovations (RECITI) 2012, July 25 – Friday July 27, 2012. Uyo-Nigeria

A Framework for Resource-Sharing Virtual LaboratoryArchitecture

Ezugwu E. Absalom1, Buhari M. Seyed2, Obiniyi A. Afolayan1, Junaidu B.Sahalu3

1. Department of MathematicsFaculty of Science, Ahmadu Bello University, Zaria-Nigeria

2. Department of Computer Science, Faculty of Science, University of Brunei Darussalam - Brunei

Darussalam3. IyaAbubakar Computer Centre,

Ahmadu Bello University, Zaria-Nigeriae-mail: {ezugwu.absalom, mibuhari,aaobiniyi,abuyusra}@gmail.com

Corresponding author: [email protected]

AbstractResearch institutions across the country continue to perform their experiments inisolation often using classic and static experimental methodologies to achieve theirrespective goals. Despite the recent advancements in the IT industries, the nations’universities and research laboratories have failed to adequately utilize these ITresources mainly due to lack of funds and human expertise. However, the currentadvances in the IT field can be applied to properly support certain complexrequirements in these domains. In other words, researchers should be assisted withconducting their complex scientific experimentation and supporting theircollaboration with other scientists irrespective of their domain and geographicallocations. The major requirements identified with all the research laboratoriesinclude, the management of large data sets, distributed collaboration support, andhigh-performance issues. This paper presents an architectural model of a virtuallaboratory framework for collaborative research. The model is open source driven,flexible and based on modern tools and technologies. This in effect will allowgeographically remote researchers with limited internal resources, access to a wealthof experimental datasets, computing facilities, and federated processed experimentalresults. The key issues to be discussed are Web-based Grid technologies, Web ServiceTechnology, and the proposed virtual laboratory architectural model. Thisarchitectural framework, besides theoretical modelling, will also provide a road mapfor future research and open questions.

Keywords: Virtual laboratory, Web services, Grid technology,Architectural framework

1. Introduction

Virtual laboratory is a heterogeneous, distributed environment, whichallows scientists across geographical locations to work on a commonproject. This environment creates much more favourable and conducive

1


atmosphere for scientists to conduct experiments with the usage of physicaldevices, perform simulations using computational resources, and initiatescommunication among researchers of similar interests over a distributednetwork environment. Technically, we could also refer to this as a group ofintegrated components that form an advanced environment, which could beused to plan and perform complex scientific applications, calledexperiments (Krakow, 2009). Experiments in this context are processes thatcombine available applications, services, tools and data into orderedactivities sequence and perform computer simulation in order to obtain newknowledge.

The main reason for building such open, flexible, scalable and configurablesystem is to allow remote access to the scientific equipment, to facilitateand accelerate research process, share knowledge among research groups andalso to manage experimental data and results. However, the system providesmultidisciplinary functionalities, which seems attractive especially to theexperimental scientists, technologists and engineers (Khetty and Xian-He,2000). Although, executing such project is quite capital intensive, theconcept of the virtual laboratory remains the long-awaited solution of thefuture for research environments with limited equipment resources.

The virtual laboratory seems the only way that will widely open the fieldof science for societal demands in education and professionalspecialization in new technologies. The realization of the proposed virtuallaboratory environment will allow scientists and engineers to work on theirprojects via remote events simulation, interpretation of experimental dataand, in some cases, run real experiments in a customized laboratory.

In the rest of the paper, we describe how virtual laboratory fits into theGrid architecture, states the anticipated goals to be achieved, presents afuturistic architectural model of this framework based on some platforms weidentified. Similarly, some of the major architectural requirements neededto successfully run distributed workspace environment are presented andfinally the advantages associated with implementing the generic distributedvirtual laboratory system are discussed.

2. Related Works

The first work on virtual laboratory comprised solely of texts and imagesof instruments, experiments, concepts, sites and people linked toexperimentalization of life (Dieriget al., 2012). Then it was also seen as aplatform for discussing experimentation in the areas of life sciences, artand technology. In recent times, two main purposes of virtual laboratorydevelopment have been identified and pursued to date. These two areas areeducation and research. The examples of educational functions ofvirtual laboratories are (Handschuh, 2012):

1. Chemistry learning and teaching by presenting simulations ofexperiments;

2. Physics presentations, such as structures and properties ofmolecules;

2


3. Familiarizing with science, including genomics and techniques appliedin biology and medicine, example. DNA microarray technology;

4. Demonstration of statistic concepts and methods.

In science, virtual laboratories are becoming more and more popular,facilitating large-scale bioinformatic studies (GenGrid, 2012 and Susumuetal., 2005) and computational tasks, for example, drug discovery (Buyya,2003), virology (ViroLab, 2012), proteomics and mass spectrometry (PubMed,2012) and other disciplines. They also enable collaboration between reallaboratories and companies (PSNC VLAB and AlmaGrid, 2012). Nowadays, due totechnological breakthrough and automation of experimentalprocedures, virtual laboratories may not only collect, store and processdata, but also perform some steps of experiments, providing access toexpensive and specialized instruments (Afsarmaneshet al., 2000). Suchapproach was applied to establish virtual laboratories connected withradio-telescope and NMR spectroscope (PNSC-NMR, 2012).

To construct and implement virtual laboratories the grid architecture isused (see section 2.2). Grid computing allows for increase of thecomputational power due to combining of multiple resources andimplementation of any application. Connecting even geographicallydistributed experimental component resources in a functional networksimplifies complex data processing and management. Virtual laboratoriesbelong to a specific representation of grid application often referred toas Remote Instrumentation Systems (Handschuhet al.,). They cooperate withother grid systems such as Globus Toolkit (GridLab, 2012). Interactivetasks are scheduled according to dynamic measurement scenario (Lawenda etal., 2004 and 2012). Processed data are stored in digital libraries(Lawendaet al., 2012). Independent of an application, each virtual laboratoryconsists of the following elements (PNSC VLAB, 2012):

1. Global access via Internet by a Web Portal: Such solution willfulfill the main condition of Virtual Laboratory.

2. Computational server: a high performance computer which can work withlarge scale simulations and data processing.

3. Databases which contain application-specific information: such asinitial simulations, bound conditions, experimental observations,client requirements or production limitations. Databases also containdistributed, application-specific resources (an example is therepositories of human genotype). The databases content should bechanged automatically. Databases could also be distributed. It shouldbe presumed that databases will contain a large amount ofinformation.

4. Scientific equipment connected with the computational networks. Forexample, it could be data from satellites, earthquake detectors, airpollution detectors, astronomical equipment.

5. Collaboration and communication tools, such as chat, audio- andvideo-conferences or tele-immersion.

6. Software: Each virtual laboratory is built on specific software whichallows for the simulating process, data analyzing or visualization.

3


Most of this software was not developed for the distributed networksolutions - it is one of the main problems in Virtual Laboratorybuilding.

2.1. State of the Art Overview

In recent times, there has been consistent innovative drive towards thedevelopment of highly sophisticated and complex IT infrastructures by thecomputational scientists, which are provided to the life sciences andnatural sciences domains. Among the most conspicuous of these advancementsis the proliferation of both web and grid technologies often deployed forbuilding robust applications in the domain of life sciences. With thecurrent demands for efficient and effective processing power forexperimental data, these technologies have moved from their staticarchitectures to more dynamic and service-oriented ones. The nextgeneration development in these technologies is coalescing towards anintegrated and unified Grid-based web services environment (Azharet al.,2008).

2.1.1. Web Service Technology

Web services technologies are among the recent breakthrough in the area ofinformation technology fields which extends the capability of themainstream web technology by promoting resources and information sharingamong devices especially in a distributed heterogeneous computingenvironment. These services could be likened to the common object models(COM) that are used to build the Windows applications.

In the Windows technology, similar services were often be rendered viatightly coupled distributed computing protocols such as the distributedcomponent object model (DCOM), common object request broker architecture(CORBA), and remote method invocation (RMI). Despite the strength of theCOM objects protocols in building specific customized applications, itstill lacked the desired flexibilities, as it is being constrained bydependencies on vendor implementations, platforms, languages, or dataencoding schemes that severely limit interoperability (Ciancettaet al.,2008). Studies and implementations have shown that applications developedas web services can easily interoperate with peer applications and this canbe attributed to the fact that the web protocols are completely vendor,platform, and language independent.

Web services are published by the Web Service Description Language (WSDL)and deployed and discovered through Universal Description, Discovery andIntegration (UDDI) protocol. These services exchange information which areusually built based on XML format by means of the Simple Object AccessProtocol (SOAP) over different computer platforms. More so, web services

4


have the capability of retaining the state of information about particulartransactions during execution period with the introduction of Web ServiceResource Framework (WSRF). Basically, these attributes associated with webservices have made them more demanding and a better choice being opted fordeveloping distributed heterogeneous web based applications and a morenecessary tool for the implementation of the proposed e-virtual libraryframework.

2.1.2. Grid Technology

The characteristics of resources needed to carry out some experiments inthe laboratory such as molecular dynamics calculations, chemical chainreaction, strength of material calculation and chemical titrations, fitswell with the concepts of the Grid. Historically, Grid has grown on top ofthe Internet, which is now available world-wide. The Internet itself beganas research project by the military. Grid comes one step next in resourcessharing – not only that information can be shared, but also processingpower, software, maybe some devices and, ofcourse, data storage capacitiesare also shareable. Another fundamental idea of the Grid is that resourcesare provided for groups of people with the same interests - VirtualOrganizations (Salnikovet al., 2009). Since its inception, the main focus ofgrid technology has been to provide platform independent global and dynamicresource-sharing service in addition to co-ordination, manageability, andhigh performance. In order to best satisfy these goals, its basicarchitecture has undergone substantial changes to accommodate otheremergent technologies.

The grid has moved from its initial static and pre-web servicearchitecture to a more dynamic Web Service Resource Framework (WSRF) basedOpen Grid Service Architecture (OGSA) (Foster, 2005). This architecturecombines existing grid standards with emerging Service OrientedArchitectures (SOAs) and web technologies in order to provide an innovativegrid architecture known as service-oriented semantic grid. The maincharacteristics of this service-oriented semantic grid would be to maintainintelligent agents that could act as software services (grid services)capable of performing well-defined operations and communicating with peerservices through uniform standard protocols (Azharet al., 2008).

3. Research Goal

This research proposes the development of a Web-based Grid platform thatleverages HPC processing environments and through which researchers,scientists and students can access their specific application domain fromtheir own remote terminals and perform experiments as if it were in aphysical laboratory. However, the target of this Design and architecturalframework is to initiate a regional software platform of a generic HPC

5


Virtual Laboratory, developed by taking into account the requirementscommon to every research laboratory within the nations’ universities, andeasy to customize by including those computation, data analysis andvisualization tools required by a specific research group, or laboratory,allowing scientists to select and evaluate efficiently and dynamically themost relevant functional and phenotypic information supporting knowledgediscoveries. Our proposed research interest is focused towards addressingthe following issues.

a. Implementing a Web-based Grid platform by utilising commoditytechnologies and open source programming tools.

b. Extending the concept of parallel processing and applying it tosolving societal problems, with special interest in distributedcomputing and federated information management methodologies.

c. Developing a distributed heterogeneous computational framework thatwill be generic enough to accommodate different Laboratoryexperimental models.

d. Design of an efficient web-based grid resource allocation managementand application scheduling algorithms.

e. Building an interactive and user friendly application interfacesusing AJAX models such as XML, JavaScript, HTTP, and XHTML. AJAX is apowerful Web development model for browser-based Web applications.

f. Building a co-working environment for developers that is used todevelop Grid-based consumable services specific to each collaborativeresearch domains.

g. Deploy the concepts of web services for application integration viathe use of a standard mechanism to be described, published,discovered, invoked and consumed by developers’ applications andservices.

4. Envisaged Generic Virtual Laboratory

The generic Virtual Lab prototype model will be composed of data analysis,and presentation infrastructures, all developed by taking into accountavailable standards, major state-of-the-art efforts and open sourcesoftware. The envisaged framework plans to fulfil the main requirements ofdesigning a flexible and configurable application suite that will offerscalability and support to specific application-oriented requirements.However, there are many design related issues that need to be incorporatedwhen designing virtual laboratory architecture such as the propermanagement of large data sets, information sharing for collaboration, anddistributed resource management (Afsarmanesh et al., 2001).

The data infrastructure is going to include a Data Warehouse that willintegrate multidisciplinary information available from several distributed

6


experimental databanks through the optical Internet. The design frameworkwill incorporate into its functionality a federated database to support thecollaboration and information sharing among database resources. The datainfrastructure is going to efficiently provide data access to theexploration and analysis units implemented in the analysis infrastructure.The presentation infrastructure is going to allow intuitive access to theintegrated data and analysis units through advanced Web interfaces.Physical experiments studied in the remote labs and specific domain-oriented applications will be used as benchmark problems to develop thesenew approaches.

4.1. Architectural Prototype Requirements

In [17] the four architectural layers for the virtual laboratory wereidentified and discussed. Similarly, in this paper we maintain the sameconcept, but with difference in general design model. The fourarchitectural layers are listed below:

a. Access Layer: This layer comprises of modules which grant access tothe virtual laboratory facilities. The module consists oflaboratories information management, teamwork environment, dynamicmeasurement scenarios and data presentation tools.

b. Grid Layer: The Grid layer allows the management of the availableGrid resource. In essence, the Grid layer provides access andmanagement services for the use of distributed computing node andtools connected to some specific laboratory devices. This layerconsists of authentication centres, global scheduling, experimentalreport archives and grid gate-way.

c. Supervisory Layer. In this layer, services are categorized inaccordance with there tasks specific types and implemented withspecified device. This layer consists of local scheduling, resourcemonitoring activities and previously used resource history/accountinformation.

d. Resource Layer. This layer is made up of resource required to executelaboratory experiments and also enable access to the efficientutilization of a pool of computational resource and visualizationsoftware managed by some local resource management systems. Thislayer comprises of laboratory apparatus, computation server andvisualization server.

The proposed architecture is presented in fig. 2. It is a conceptual way toshow abstract layers of the virtual laboratory; however the realarchitecture is far more complex than what is shown below. The mostimportant thing is that the proposed model of virtual laboratoryarchitecture is designed to be both scalable and extensible. Scalable interms of enabling a large number of research groups to collaboratively

7


setup, execute and discuss about their experiments and its outcomes viatools or software that operate in real-time bases. In the order hand,extensible in terms of enabling the addition of new modules, changing ormodifying existing module without having to recompile the entire processagain.

Despite the aforesaid features associated with building a robust virtuallaboratory prototype model, it is still good to be mindful of the fact thatmost laboratory experimental setups are specific and differ by the wayexperiments are being carried out in them. Therefore, we suggest thepossibilities of building just a new virtual laboratory only by definingnecessary modules and adapting it to the surrounding environment. Specificdomain or application designers need not care much about the systemarchitecture, but rather, emphases should be focused on building laboratorygroups, assuring collaboration between apparatus and the existing virtuallaboratory modules

Figure 1: Virtual Laboratory Layered Architecture

4.2. System Workflow Plan

Figure 2 is a highlight of the research areas which are of paramountimportance to the proposed study. The framework design model will majorlyconcentrate on service areas such as the computational services, dataservices, and the experiment developer, a service which is the assignedrole of the author. The system workflow is modelled as follows; theexperiment developer (programmer) first plan, develop and executes his planbased on the computational services available. Experiments are usuallycarried out based on available data gathered from collaborating databases,which is to be made possible by the data service platform. The prepared

8


experimental suits are nothing more than laboratory emulators and are meantto work akin to the physical laboratory apparatus. These emulators will bepublished via the web services platform from which laboratory users such asscientists, researchers, technologists and students will use to performtheir experiments. They should be able to know which data to work on and toperform expected analyses and visualization, as a means of interpreting theresults of such experiment.

Figure 2: The Framework of the Virtual Laboratory Workflow

4.3. Virtual Laboratory Resource Centre

Presented in Figure 3 is Virtual Laboratory Resource Centre (VLRC) withdistributed heterogeneous resources. A VLRC consists of n virtual siteswhich are connected to the Grid via a wide area network with bandwidthstrength commensurate to the site’s local bandwidth strength. A virtualsite may consist of multiple physical sites if they are interconnected by ahigh bandwidth network. Each site consists of computational servers,laboratory devices, storage systems, visualization servers and software forexperiment execution. A virtual site can be thought of as a “regionalresources centre”, a composite object containing a number of data servers,processing nodes and software for laboratory experiment execution where allare connected to a local area network.

9


Figure 3: Virtual Laboratory Resource Centre

5. Research Benefits

Collaborating scientists, technologists and engineers across Nigerianuniversities and industries that will be involved in this project stands torip the benefit associated with distributed and dynamic informationsharing-resource portal with friendly user interfaces. The successiveimplementation of this scalable virtual high performance environment willbe seen as community knowledge store, whose main focus is on broadening thespectrum of experimental analysis by facilitating new discoveries withinmultidisciplinary domain, as well as providing a more universal perspectiveof research which would spur the process of revealing the unifyingprinciples in sciences.

The proposed virtual laboratory will also make available those vital, butunaffordable advanced research equipments such as complex HPC tools thatare required to store, manage and analyse enormous volume of data generatedin different research laboratories across the country. Most importantly,the work will aid researchers and specialist to identify new methods ofproblem solving techniques with broader perspective in scientificreasoning. In the long term, these will allow the design and optimization

10


of scientific production processes in a reliable, predictive andquantitative way.

6. Discussion and Conclusion

The modelling of a laboratory framework is a very difficult task toperform, especially by trying to define a generic framework that will beappropriate to the different laboratory equipments being used in variousresearch domains. In similar vein, finding the solution that will meet thedemands of most research groups and also permit experiments to be carriedout using different apparatus will consume a lot of time, cash, resourcesand energy. However, achieving these goals is worth the effort, as it willattract a lot of advantages such as reduction in resources that will berequired to establish a new laboratory.

The actualization of the virtual laboratory environments to its fullestcapacity will allow researchers to effectively and efficiently work ontheir projects via remote access to more advanced laboratory apparatus andcomputational tools that will enable them interpret their experimental dataand in some cases run real experiments in a customized laboratory. Thesuccessful integration of the web and grid technologies, and the optimalutilization of modern advancement in information technologies is a perfectplatform for this project development.

References

Afsarmanesh H., Benadbelkader A., Kaletas E. C., and Hertzberger L.O,"Toward aMulti-layer Architecture for Scientific Virtual Laboratories." In: Bubak M.,Afsarmanesh H., Williams R., HertbergerB., (Eds), Proc. Int. Conf. HighPerformance Computing and Networking Amsterdam, May 8-10, 2000, Lecture Notein Computer Science 1823, 162-179, Springer, 2000

Afsarmanesh H., Kaletas E. C., Benabdelkader A., Garita C., Hertzberger L. O., “Areference architecture for scientific virtual laboratories”, Future GenerationComputer System 17, 2001 999-1008. Elsevier.

AlmaGrid: site accessed on 12-Mar-12 athttp://www.clustervision.com/pr_almagrid_uk.php

Azhar A., Barthel D., Lukasiak P., Blazewicz P., Krasnogor N., “Web and GridTechnology in Bioinformatics, Computational and Systems Biology”: A Review,Current Bioinformatics 2008, 3, 000-000, Benthham Science Publishers Ltd.

Buyya R, Branson K, Giddy J, Abromson D. The Virtual Laboratory: enabling molecularmodelling for drug design on the World Wide Grid. Concurrency Computat: Practand Exper 2003; 15:1-25.

Ciancetta F., D’Apice B., Gallo D., Landi C., “Plug-N-Play Smart Sensor Networkwith Dynamic Web Service”, IEEE Transactions on Instrumentation andMeasurement, vol. 57, no. 10, October 2008.

Dierig S., Schmidgen H., Kursell J. “The Virtual Laboratory. Essays and Resourceson the Experimentalization of Life”, website accessed on 29-Jun-12 athttp://vlp.mpiwg-berlin.mpg.de/index_html.

11


Foster I., Service-oriented science. Science 2005; 308: 814-17.Genegrid—virtual bioinformatics laboratory.website accessed on 28-Jun-12 at

http://www.qub.ac.uk/escience/dev/article.php, sectionprojects.GridLab: A Grid Toolkit and Testbed, Website page accessed on 07-Jan-12 at

http://www.gridlab.org/Handschuh L., “Virtual laboratories as the new approach to genomic experiments”,

website accessed on 15-Jan-12 athttp://lib.bioinfo.pl/courses/question_list/515.

Handschuh L., Lawenda M., Stepniak P., Figlerowicz M., Stroinski M., Weglarz J.,“New Approach to Genomics Experiments Taking Advantage of Virtual LaboratorySystem. ” web site page accessed athttp://www.man.poznan.pl/cmst/2009/_V_15_1/03-Handschuh_L_31-40.pdf.

Khetty D., Xian-He S., “A windows-NT virtual collaboratory for technicalcomputing”, Fifth NASA National Symposium on Large-Scale Analysis, Design andIntelligent-Synthesis Environments. Williamsburg, VA, USA. 12-15 Oct. 1999.Elsevier. Advances in Engineering Software, vol.31.no.8-9, Aug.-Sept. 2000.

Krakow, AGH University of Science and Technology in Krakow, “Bioinformaticsapplications in the Virtual Laboratory”, 2009.

Lawenda M., Meyer N., Rajtar T., Okon M., Stoklosa D., Kaliszan D., Stroinski M.,“Generalization aspects in the virtual laboratory system.” PoznanSupercomputer and Networking System, web site page accessed on 07-Jan-12 athttp://www.psnc.pl/

Lawenda M., Meyer N., Rajtar T., Okon M., Stoklosa D., Kaliszan D., Stroinski M.,“Job workflow in the Virtual Laboratory.” Global Grid Forum 10, Berlin,Germany Grid Workflow Workshop. March 9, 2004

PSNC VLAB, “Virtual laboratory definition and structure”, website accessed on 28-Jun-12 at http://vlab.psnc.pl/vldef.html.

PSNC-NMR.“Radiotelescope and NMR spectroscope in VLab.” website accessed on 15-Jun-12 at http://vlab.psnc.pl/vlworks.html.

PubMed: Web site accessed on 12-Jun-12 athttp://www.ncbi.nlm.nih.gov/pubmed/19358578

Salnikov A. O., Sliusar I. A., Sudakov O. O., Savytskyi O.V., Kornelyuk A.I.,“MolDynGrid Virtual Laboratory as a part of Ukrainian Academic Gridinfrastructure.” IEEE International Workshop on Intelligent Data Acquisitionand Advanced Computing Systems: Technology and Applications 21-23 September2009, Rende (Cosenza), Italy.

Susumu D, Kazutoshi F, Hideo M, HarukiNShinji S. “An Empirical Study of GridApplications in Life Science: lessons learnt from Biogridproject in Japan”,International Journalof Information Technology 2005; 11: 16-28.

ViroLab: Website page accessed on 07-Jan-12 at http://virolab.cyfronet.pl/trac/vlvl

12

A Framework for Resource-Sharing Virtual Laboratory Architecture

Documents