Titulo Estágio
CloudAid 2
Área Tecnológica
Sistemas de Informação
Local do Estágio
DEI-FCTUC
Enquadramento
Cloud Computing considers software and hardware as utilities (dealing with Anything-as-a-Service, such as Platform-as-a-Service, Infrastructure-as-a-Service or Software-as-a-Service) and brings new challenges, namely to decision making when selecting a supplier of a Cloud service or an aggregation [1] of these to compose another added value Cloud service. The first companies focusing on service aggregation and integration are emerging. For example, runmyprocess.com and jitterbit.com enable to integrate various cloud services as one orchestration. Nonetheless, they do not provide any solution to assist the decision making when selecting cloud services. In a previous master thesis the CloudAid system [2] was developed to suggest the aggregation of cloud services following user requirements. CloudAid proposes the aggregation of cloud services such as those of [url=http://aws.amazon.com/ec2/]Amazon EC2[/url] and [url=https://www.heroku.com/]Heroku[/url], based on functional requirements and non-functional ones, such as the service price. These services are modeled with [url=http://www.linked-usdl.org/]Linked USDL[/url] [3], which is a semantic service description language based in USDL [4, 5]. Linked USDL is built upon the Linked Data (LD) principles, which enable to create typed links between data from different Web sources resulting in a web of things in the world, described by data on the Web. Since Linked USDL provides a business perspective description of services combined with semantics, end-users will achieve a higher precision and recall during their search for services.
The selection and aggregation of Cloud services is a multi-criteria decision problem [6] because there are several user requirements and preferences to match with the characteristics of those services. Besides the service price, the decision maker has to take into account several Cloud service characteristics such as security, legal constraints, availability, storage and processing capacities, to name a few. As so, the CloudAid system describes the decision problem in a generic model, using MultiCriteria Decision Analysis and the XMCDA data standard [7]. Applying this model, CloudAid implements two examples of multi-criteria decision algorithms that are SAW (Simple Additive Weighting) and AHP (Analytic Hierarchy Process), hence to provide aggregations of Cloud services that may help the decision maker to do an assessment on the best services set.
CloudAid produced promising results and in the current master thesis it is intended to deepen the development of this system.
This MSc thesis work may involve collaboration with SAP Research and also researchers of the Faculty of Electrical Engineering and Computing of the [url=www.fer.unizg.hr]University of Zagreb [/url].
References
[1] Cloud aggregator, December 2011, http://searchcloudprovider.techtarget.com/definition/cloud-aggregator
[2] CloudAid, Estágio de mestrado de Jorge Araújo, 2012/2013.
[3] C. Pedrinaci and J. Domingue, Toward the Next Wave of Services: Linked Services for the Web of Data. In Journal of Universal Computer Science, vol. 16, no. 3, pp. 1694 -1719, 2010, http://oro.open.ac.uk/23073/1/services-for-the-web-of-data.pdf
[4] Cardoso, J.; Barros, A.; May, N. and Kylau, U. Towards a Unified Service Description Language for the Internet of Services: Requirements and First Developments. In IEEE International Conference on Services Computing, IEEE Computer Society Press, Florida, USA, 2010.
[5] Cardoso, J.; Winkler, M. and Voigt, K. A Service Description Language for the Internet of Services. In First International Symposium on Services Science (ISSS'09), Leipzig, Germany, ISBN: 978-3-8325-2169-1, 2009.
[6] J. Figueira, S. Greco, and M. Ehrgott. Multiple Criteria Decision Analysis: State of the Art Surveys. Springer Verlag, Boston, Dordrecht, London, 2005.
[7] Decision Deck project, http://www.decision-deck.org/xmcda/
Objetivo
The main objective of this MSc thesis work is to enrich the CloudAid system, by adding new features to the existing modules and developing new ones. This will produce the CloudAid 2 system. These new features include, but are not limited to:
• Design and implementation of a user-friendly and ergonomic visual graphical interface. The current user interface of the CloudAid system is textual. It is intended to apply ergonomic design guidelines to create a visual graphical human interface that facilitates the user interaction with the CloudAid system and its usability;
• Implementation of a triplestore of Cloud service descriptions. This involves also applying Web crawling and scraping techniques to collect Cloud Services and transform these in Linked USDL service representations to feed the triplestore of services;
• To take into account other Cloud service characteristics (some of these were defined in a CloudTaxonomy, during the previous master thesis) - such as service availability, service dependencies, service quality and service user opinions and recommendations - when processing the aggregation of services. The CloudTaxonomy will be extended with the new cloud service characteristics that were not included in the previous version. This means the CloudAid 2 system should consider other Cloud service non-functional characteristics and related information, such as monitored service availability and cloud service user opinions and recommendations that can help the decision maker. This step requires capturing these service characteristics, representing them in Linked USDL service description and use these when comparing service descriptions with the list of user requirements to recommend the best cloud service solutions;
• Implementation and testing of other multi-criteria decision methods to compare service descriptions with the list of user requirements and recommend the best cloud service aggregations. Examples are outranking or fuzzy methods, allowing to deal with incomplete information and incomparability;
• Include features that will improve the tool ease of use and user experience, such as storing service request history to improve further user interaction (namely, prompt the user to modify its requirements of service request if no service is found for the user previous requirements), grouping service alternatives (that have the same values for the user defined criteria) and allowing the user to define an indifference threshold enabling the filtering and grouping of similar alternatives;
• Implementation of the automatic calculation of dynamic characteristic functions namely price (that were defined with Linked USDL properties) using the SPIN SPARQL Inference Notation. SPIN [8] is a standard to represent rules and constraints that enable the calculation of the value of a property based on other properties and the isolation of a set of rules to be executed under certain conditions.
For this, this MSc thesis work has the four main following sub-objectives:
1. Analysis phase: studying of the CloudAid system and reading of existing documentation. Detailed specification of the CloudAid 2 system requirements and its prioritization. Reading and analysis of existing methods and concepts, such as multi-criteria decision aid and SPIN.
2. Design phase: specification of the architecture of the new modules and new features that satisfy the CloudAid 2 requirements. This specification has to take into account the integration with the existing CloudAid system.
3. Implementation phase: programming will be used to develop and implement the CloudAid 2 system designed in the previous phase;
4. Verification and validation phases: test cases will be defined and implemented to evaluate whether the CloudAid 2 system fully or partially satisfies the outlined requirements. After testing this system, defects, if found, will be logged and will enable to correct the implementation. A system user manual will also be prepared.
[8] SPIN SPARQL Inference Notation, W3C submission, http://spinrdf.org
Plano de Trabalhos - Semestre 1
According to the objectives explained in the previous section, this MSc thesis divides into the following tasks, each one with concrete outputs (deliverables):
Task T1: Analysis of the CloudAid system and detailed specification of the CloudAid 2 system.
- When: September-October-November 2013.
- Description: study of the CloudAid system and reading of existing documentation. Detailed specification of the CloudAid 2 system requirements and its prioritization. Reading and analysis of existing methods and concepts, such as multi-criteria decision aid and SPIN.
- Deliverable: Report indicating the detailed specification of the CloudAid 2 system requirements and its prioritization.
Task T2: CloudAid 2 architecture and design.
- When: November-December 2013.
- Description: Specification of the architecture of the new modules and new features that satisfy the CloudAid 2 requirements. This specification has to take into account the integration with the existing CloudAid system.
- Deliverable: Report of the architecture of the CloudAid 2 system.
Task T3: Mid-term report writing.
- When: December-January 2013-2014.
- Description: Write the report for the mid-term MSc evaluation.
- Deliverable: Mid-term MSc thesis report.
Plano de Trabalhos - Semestre 2
Task T4: Implementation of the CloudAid 2 system.
- When: February-May 2014.
- Description: programming to develop and implement the CloudAid 2 system designed in the Task T2;
- Deliverable: Prototype of the CloudAid 2 system and corresponding documentation.
Task T5: Tests and evaluation of the CloudAid 2 system.
- When: May - June 2014.
- Description: Definition and implementation of test cases enabling to evaluate whether the CloudAid 2 system fully or partially satisfies the outlined requirements. The tests should be exhaustively executed and reported. If defects are found during this testing, these will be logged and will enable to correct the implementation.
- Deliverables: An evaluation report of CloudAid 2 testing and the system user manual.
Task T6: Final MSc thesis report writing.
- When: May-July 2014.
- Description: Write final MSc thesis.
- Deliverable: MSc thesis.
Condições
This work will be carried out at DEI/Universidade de Coimbra within the Information System Group (ISG). A suitable space will be made available to the student. Meetings will be held regularly (every week) with several students working in the [url=http://genssiz.dei.uc.pt]Genssiz Center for Services Systems Research[/url]. A FCT scholarship is available.
Observações
This MSc thesis work will be jointly supervised by the PhD Catarina Ferreira da Silva (CISUC, cferr@dei.uc.pt), the Professor Jorge Cardoso (DEI, jcardoso@dei.uc.pt, 239790051) and the Professor Paulo Melo (FE, pmelo@fe.uc.pt).
Orientador
Catarina Ferreira da Silva / Jorge Cardoso / Paulo Melo
cferr@dei.uc.pt 📩