Self-organization and SOA for networked enterprises

Introduction

The needs for exibility and globalisation force enterprises to both decentralise their activities and continuously (re)structure their network of relationships, regarding their productive processes, their \supply chain", and their design and innovation processes.

This demands for new innovative technological solutions, which must support unprecedented levels of dynamism. In particular, the overall architecture of the information system must evolve as requirements and context change. Typical cases of context changes occur when new services become available, existing services become less competitive, or even when they are discontinued. As soon as deviations from the desired or acceptable quality of service are discovered, or new opportunities are devised, the information system should be able to recon gure itself in an autonomic manner. SOAs naturally t in this context as they inherently support the concept of service, which represents one of the typical items organizations exchange, and also because SOAs are suitable to be equipped with self-adaptation mechanisms that enable them to evolve on the y.

In this context, our research work is focusing on addressing the challenges presented below.

Recognizing network evolutions and actuating the needed actions

Networked enterprises usually are created on the basis of some need and trust relationships between stakeholders. Of course, networks that are created in the rst place are not necessarily optimal. Also their performance can evolve over time. Thus, it is important to equip the underlying SOA infrastructure with monitoring mechanisms able not only to identify issues at the technical level, but also to spot potential places for improvement and optimization of the network performance. Consider, for example, a simple network where there is a Service Provider and a User. The User needs a service and the Provider provides the service to the User for a fee. This network works quite well, but it does not provide to the User any knowledge about the reliability of the Provider, the possible mismatches between the advertised quality for that service and the actual one, and the correct pricing. Thus, it may happen that the lack of knowledge for the User with respect to the existence of competing providers and to the reputation of the known ones may result in a reduced satisfaction or in paying more for having less. In this speci c situation, the introduction of a third-party broker could improve the situation and allow the User to receive the required feedback and also to extend the number of its Providers [ 4 ]. While the decision on how to recon gure the service network is intrinsically a task for humans, still the IT infrastructure can help in identifying possible recon gurations as a result of a continuous monitoring and analysis activity. Monitoring should be built in a recon gurable way so that the data to be gathered are decided dynamically, depending on the situation of the network. We imagine experts in the service network domain identifying the relevant data and properly instructing the monitoring infrastructure through a simple to use, still rich language. Part of the data analysis tasks could be automated as well through the de nition of proper invariants and predicates that should be checked on the basis of the data acquired by monitoring. 3

Enabling self-evolution of the underlying SOA infrastructure

Dynamic evolution of the networked enterprises clearly results in the need for evolving the SOA infrastructure as well. In the literature, various approaches have been identi ed, but most of them are still focusing on the speci c mechanisms to use more than on the de nition of a rigorous design approach that leads to a SOA system able to evolve at runtime in a controlled way. In [1{3] we propose some preliminary approaches that are currently being extended and validated. Figure 1 shows the life-cycle we have identi ed for adaptable SOA systems. It highlights the fact that adaptation happens at runtime and goes through various phases (see the left hand side of the gure) where activities devoted to the identi cation of adaptation needs and strategies are followed by the one devoted to the execution of the adaptation, which makes the system returning back in its operation phase in an evolved con guration. Even though we have contributed to the identi cation of these phases and are working on clarifying their purposes, still speci c techniques and approaches are needed to allow designers to design the system for adaptation and evolution. 4

On the

y revision of SLAs In [ 5 ] we have presented a framework that supports design-time and runtime negotiation of SLAs in a SOA context where the negotiation participants are essentially service providers and service consumers. The framework also o ers the possibility to replace (part of) these participants with software agents that act on behalf of them and to design new negotiation processes, possibly including multi-parties, which are then interpreted by the runtime infrastructure. Di erently from many works in literature that support speci c negotiation processes, Monitored events

Identify adaptation needs

Monitored properties

Context Adaptation requirements Suitable strategies

Identify adaptation strategy

Strategy instance

Enact adaptation

Adaptation mechanisms Decide between adaptation and evolution Run-time monitoring

Operation & management

Deployment & provisioning

Requirements

Engineering &

Design

Deploy adaptation and monitoring mechanisms Deployment-time adaptation

Early Requirements

Engineering

Define adaptation and monitoring requirements Design for monitoring and adaptation

Construction

Construction of monitors and adaptation mechanisms actions

Design artifacts

our framework can be tailored depending on the multiplicity, work ow, protocol, and decision model that t a speci c application domain. The evolution in the networked enterprise or in the underlying IT infrastructure may determine the need for revising the SLAs established between the involved stakeholders. Consequently, approaches to identify needs for revision and to renegotiate SLAs are required. 5

Conclusions

In this position paper we shortly describe the research challenges we are currently focusing on. These concern, in particular, the need for identifying and applying potential changes in a network of enterprises, the need for supporting self-evolution as a result of changes in the network, the need for revising during the operation process the already agreed SLA. We envisage that new innovative technological solutions in these areas are necessary for addressing the service requirements of networked enterprises in the Future Internet.

Acknowledgements

Our research is partially funded by the European Community, Seventh Framework Programme FP7/2007-2013 under grant agreements 215483 (S-Cube), 216556 (SLA@SOI) and 215219 (SOA4All).

References

1. A. Bucchiarone , C. Cappiello , E. di Nitto , R. Kazhamiakin , V. Mazza , and M. Pistore . Design for adaptation of Service-Based applications: Main issues and requirements . In International Workshop on Engineering Service-Oriented Applications , 2009 .

2. A. Bucchiarone , R. Kazhamiakin , C. Cappiello , E. Di Nitto , and V. Mazza . A context-driven adaptation process for service-based applications . In PESOS 2010 { 2nd International Workshop on Principles of Engineering Service-Oriented Systems , Cape Town, South Africa, 2010 .

3. L. Cavallaro , E. Di Nitto , C. Furia , and M. Pradella . A Tile-based Approach for Self-assembling Service Compositions . In In Proceedings of International Conference on Engineering of Complex Computer Systems , 2010 .

4. D. Dubois , C. Nikolaou , and M. Voskakis . A model transformation for increasing value in service networks through intangible value exchanges . In International Conference on Service Science, ICSS2010 , 2010 .

5. E. D. Nitto , M. D. Penta , A. Gambi , G. Ripa, and M. L. Villani . Negotiation of service level agreements: An architecture and a search-based approach . In B. J. Kramer, K.-J. Lin , and P. Narasimhan, editors, ICSOC , volume 4749 of Lecture Notes in Computer Science, pages 295 { 306 . Springer, 2007 .