next up previous contents
Next: Sakai Evaluation including CHEF, Up: Role of Portals in Previous: A Service Oriented Architecture   Contents

Technical Framework

This section briefly outlines technical aspects of a VRE which may indicate the scope of work required. This discussion arises directly from the identification of an SOA as the preferred architecture for the VRE and the community input which has led to the workshop recommendations. It is easy to be all-inclusive in a theoretical approach, but a more carefully chosen set must be suggested for pragmatic reasons. The following sections therefore attempt to suggest an initial preferred set.

Frameworks/ Interfaces

Following discussions among interface providers, such as those writing 2nd Generation Portals and web services based middleware, we suggest adopting the following standards:

$\bullet$ JSR-168 (Java standard for portlet interface);
$\bullet$ Web Services (WSRP, WSDL, WS-I) which may extend to such things as WSRF in future (SOAP over HTTP is mainly implied as a request/response protocol but other delivery protocols may be needed, e.g. for large datasets, see below);

Frameworks should therefore be able to support these interfaces, e.g. in the case of portals Websphere, uPortal, Jetspeed, GridSphere etc. are possible candidates. For further discussion on JSR-168 and WSRP see [28]. There will be multiple solutions for access to the VRE and standards and inter-operability are key issues. In addition to frameworks such as portals, services must be able to be incorporated into heritage applications and GUI interfaces such as those developed and maintained by the long-standing Collaborative Computational Projects (CCPs), which are a unique contributor to UK research output.

Generic Services

These generic services are an abstraction of actual services, some of which might be made available via the VRE. Appendix H provides a fine-grained, but still evolving, list of services from areas such as:

$\bullet$ AAA, Grid Services, Semantic Services, Resource Discovery, etc.
$\bullet$ E-Collaboration;
$\bullet$ Support and Management Services.

This is an attempt at an initial classification and description of these services based on work of other groups such as JCLT, ETF and GGF. Further work is required to identify how services can be broken into methods (functionality) and if appropriate where existing middleware provides such functionality. See http://www.grids.ac.uk/ETF/public/WebServices/classes.html.

Real Resources

Services must clearly be mapped onto real resources to be of value. In Appendix G we have listed some resources and services already available to the UK academic research community which could be accessible. This list is also not exhaustive, but we should identify the main resources and facilities which are likely to be available via the VRE. It may be the case that some of these resources need extra effort to provide interfaces or full availability. Gaps may also be identified as we analyse these resources and services.

Requirements and Use Cases

Development of requirements is an ongoing process and the awareness days and demonstrations which have taken place so far have informed this document. Following further demonstrations of VRE prototypes additional requirements will emerge as people change their ways of working, just as the invention of the elevator by Otis made the upper floors of skyscrapers into desirable real estate.

As an example of how requirements could be used to inform the list of services and identify workflows we firstly give a very simple use case:

[frame=single]
User specifies required topic;
Uses search to discover location;
Uses authentication and authorisation to access source;
Downloads material using appropriate protocol;
Interacts with it using appropriate tool(s);
Does research;
Writes paper;
Publishes content;
Attends international conference.

Increasingly more complex and sophisticated use cases can be devised, and this should be done before large-scale deployment. Firstly we show a single-user case combining access to an HPC application on a remote resource with transfer of a large dataset from a data centre.

[frame=single]
User logs in to VRE server
   - Login is based on e-mail and password
   - Login includes using X.509 certificates for Grid resources
   - Use Shibboleth or whatever emerges from JISC security middleware
      projects (future work)
User session established/ managed/ logged/ preferences selected
User selects a "Research Service Package" on the tools menu
   - Could involve a number of underlying tools/ services mediated by
      workflow to guide user through a complex procedure
User clicks "help" button or is provided with guidance in another form
Need to invest in "accessibility" configurations
User performs a query to list available services/ information sources
      with particular search criteria and semantic support
Service query is sent to Universal Description, Discovery &
      Integration (UDDI) registry
   - Enables dynamic invocation of Grid service
   - Service providers can edit, delete or modify published grid services
      through UDDI interface
   - UDDI returns a list of requested services and a URL describing on how
      to invoke the grid service
Information/ data query sent to cross-search broker and translated to
      use appropriate metadata via an information service
   - URL returned pointing to appropriate sources
   - Translation of format may be required through another service
   - Authorisation etc. with remote database  la OGSA-DAI
   - Protocol for download chosen, e.g. GridFTP
Clicking on the URL will generate a Web Services Description Language
      (WSDL) file or initiate download by pre-defined method
WSDL converted "on the fly" to a web form and presented to the user
User enters the appropriate additional details and submits the job
   - Data transmitted to host system
   - Job submission done using Web services and job logged
   - Monitoring of job done for future inspection
   - Notification of completion or other "event" by e-mail or message board
   - Collection of output, visualisation etc.
   - Repeat job
   - Send output to a different task in a pipeline

Finally an example of using a VRE for collaborative working between two or more users.

These steps could be part of a more complex workflow which could be modified with different data/ components and/ or repeated.

User has option of saving, deleting, printing or visualising the output and making decisions during the workflow session.

Registry of services and data needed and tools to query, find and consume them.

Security (authentication), confidentiality, privacy, accessibility and authorisation mechanisms essential.

Need appropriate presentation to user in "familiar" environment, e.g. Web portal.

[frame=single]
User logs on as above
   - Private session managed as above
Determines if "peers" are present or if there is a hierarchical activity group
Asks to join group and is admitted
   - Peer session established and managed when more than one participant
Discourse with or share material with group
Use collaboration tools:
   - Whiteboards, visualisation, etc.
   - Chat, messaging, resource box
   - Controlled vocabulary
   - Decision support
   - Support for inter-disciplinary research
In research this is an important element of planning and project management

Protocols/ Standards

To make use of resources within the VRE additional protocols and standards are required over and above the access protocols identified above (e.g. Web services). Obvious requirements include the transfer of large quantities of data between data stores and computers, which requires optimal utilisation of bandwidth via a tuned FTP service. Search and content feeds also amplify the requirements. The following list may need to be extended, but should be contained to a manageable set implimented within the VRE and able to be consumed by deliery mechanisms such as portlets:

$\bullet$ Protocol examples: FTP, VIC, RAT, Z39.50, SRW/ SRU, OAI-PMH, RSS, OpenURL;
$\bullet$ Standards examples: IMS, Dublin Core, LOM, OAI, OKI, etc. including those as appropriate from W3C, IETF, OASIS, GGF.

Criteria for Success

In accordance with the discussion set out in the text, we suggest that a successful VRE would:

Call for Proposals

The discussion summarised above led to a Call for Proposals by the JISC JCSR, circular number JCSR(04)/05. This aimed to spend £ 3.2M of money from HEFCE on demonstrators, tools and components for a UK VRE. The call deadline was 13th August and the proposal evaluation was completed on 10th September and endorsed on 20th September 2004. The successful proposals using Sakai and related portal technology were:

Project Framework Community
Sakai Demonstrator Sakai To address the requirement for a single point of access to a comprehensive set of Grid and collaboration services in a VRE.
Integrative Biology OGCE To develop a VRE demonstrator to investigate the utility of existing collaborative frameworks to support the entire research process of a large-scale, international research consortium.
Silchester Roman Town Sakai To develop a system to facilitate rapidly developing and iterative archaeological research by synchronising the three processes of gathering information, co-ordinating expertise, and managing the resulting body of data.
ELVI SunGard SCT To produce and demonstrate a practical framework for the effective deployment of a generic VRE.
EVIE Bodington To test the integration and deployment of key existing software components within a portal framework.
ISME GridSphere and Access Grid To develop and refine the experimental steering process, shared workspace and distributed visualisation into a VRE making them deployable by dispersed teams of instrument scientists, material scientists and engineers in a transparent and robust manner.
IUGO   To develop a proof of concept system to enable the integration of Web-based content (and references to non Web-based content), related to individual conferences and individual sessions within conferences, thus providing a means to provide far greater benefit to the wider research community than is currently available from conference attendance.
Social Science WSRP, Sakai, DSpace To provide and support the Sakai platform for large, distributed social sciences research projects. The focus is on activities to investigate the needs of the TLRP researchers and to evaluate the extent to which Sakai meets those needs.
Politics Sakai, Access Grid To develop a virtual research and research skills development environment, capable of expansion and of facilitating multiple participation in the rapidly evolving field of the history of political discourse.
GROWL C Web services library To build upon the existing prototype GROWL library to produce a truly lightweight extensible toolkit which complements other solutions.
Humanities Sakai To build a VRE for the Humanities by investigating how Humanities research can benefit from ICT and by constructing demonstrators in specific fields.
Access Grid VRE Access Grid Extend the Access Grid's VRE infrastructure with new collaboration functionalities from the CoAKTinG project.
Cheshire VRE Sakai and Cheshire III To identify and integrate a number of tools and technologies for the data grid and digital library communities which will support collaborative e-Research across institutions and domains using Sakai.
CSAGE Access Grid To construct and use semi-immersive stereoscopic facilities to create an increased level of ``presence'' within the Access Grid environment and the recording of the performance, in the same format, within a framework playable by a larger community base as and when required.
CORE Chandler and Web services To provide integrated computer support across the research and educational cycles, because these activities are intrinsically coupled as a part of the requirements of the surgeon's continuing professional development.

The developers in all 15 fuded JISC VRE projects, whether or not they are using Sakai, have been invited to inform each other on their development and use of technology and tools, especially if they can be shared to increase the overall value of the VRE programme. To help this process we have set up collaboration tools as follows:

Sakai Portal Worksite: http://collab.sakaiproject.org under ``eResearch''. You need to register with the portal and then inform Rob Allan r.j.allan@dl.ac.uk who will enable access.

e-Research Wiki: http://www.grids.ac.uk/eResearch. Self register. The above information is taken from this Wiki.

e-Research Mail list: e-Research@jiscmail.ac.uk. Self register.

The Wiki and mail list are linked into the Portal worksite.


next up previous contents
Next: Sakai Evaluation including CHEF, Up: Role of Portals in Previous: A Service Oriented Architecture   Contents
Rob Allan 2005-05-09