Recovery of Faulty Web Applications - Essay Example

 Recovery of Faulty Web Applications through Service Discovery SUMMARY OF THE PAPER The paper details how self healing web services can be incorporated into the architecture of the systems that are hosting the web services. As the title of the paper states, the recovery of the faulty web applications can be done using the healing properties of the web services by using the WS-Diamond type of software. This software “aims at realizing a set of tools that can be plugged into existing Web service environments and provides self-healing features during the Web service life cycle” (Fugini, 2006). As the author’s state at the beginning of the paper, “The goal of this paper is to present an architecture for self-healing Web services and applications, which contains modules for the detection, diagnosis, and repair of faults. Our goal is to show how faults in Web application can be repaired using repair actions that include switching to a substitute service, by searching it in UDDI registries, based on similarity criteria. In particular, we focus on healing mechanisms based both on service selection and substitution and on addition of new services in composed processes to support self-healing functionalities” (Fugini, 2006). In these days of e-commerce transactions where much of the commercial activity is carried out over the web, it becomes imperative for developers of e-commerce applications to factor in the possibility of failure of one leg of the transaction or failure arising out of lack of connectivity or other issues in the midst of the transaction. Hence, there is a need to correct the transaction quickly and repair and restore the system to the original state without loss of information. Since, most of these transactions involve money there is a need to ensure that there is no slippage and hence lossless transactions have to be the norm of the site. PROBLEM ADDRESSED AND RESEARCH CHALLENGES As outlined above, the research problem that is being addressed is the recovery of the system arising out of a failure and ensuring that the system recovers without causing delivery failures. The authors attempt to address this research question by proposing alternative network architectures and system architectures aimed at solving the problem of recovery after system loss. The authors aim to address the issue of faulty transactions as well. Hence, the research problem being addressed can be summed up as how to recover the system in case of failure and how to avoid faulty transactions from playing havoc with the business processes of the users. The authors go on describe the types of system faults and that range from internal data faults to actor faults. These are the web-application related faults. A further category of faults would be the web-service execution faults and web-service co-ordination faults. The research challenges in the area of self healing system architectures arise from the way in which the web services have to be configured for disaster recovery and system failure. It is by no means a simple task and the implications for system recovery and self healing of the systems lie in the way in which the layered architecture is presented as an alternative. The questions posed by the research challenges have been handled well by the authors. MAIN CONTRIBUTION OF THE PAPER The main contribution of the paper is in the way in which the authors try to address the issue of faulty web services and faulty transactions arising out of system failures by proposing alternative architectures that would minimize, if not reduce, the impact of such failures on the way in which the transactions can impact the usage of the software. The authors do a good job in describing how a self healing architecture works and how it can be used to heal the system after failure. Apart from the aspect of faulty transaction repair and healing of the system, the other main contribution of the paper is towards specifying “reference architecture for faults treatment and a classification of faults have been given, together with a set of strategies for recovery. Fault occurring during Web service and Web application execution have been studied and discussed within a proposed architecture where faults detection and interpretation for repair requiring search of substitutive services have been presented” (Fugini, 2006). STRONG ASPECTS OF THE PAPER The strength of the paper lies in its ability to provide solutions to the vexed issue of faulty transactions that are the bane of any e-commerce application. As outlined in the previous sections, the fact that there are monetary exchanges in these transactions makes the systems hosting these transactions zero-tolerant towards faulty transactions. The paper has given a roadmap for achieving this goal of making the systems resistant to faults and has provided solutions for the same. The other strength of the paper is its innovative use of system models and explanation of faults along with the classification of the same to provide for an up to date model of how systems can recover along with how they can heal themselves in case of faulty transactions. WEAKNESSES AND SUGGESTIONS FOR IMPROVEMENTS One of the weaknesses of the paper is the absence of data recovery strategies. Since transactions are composed of data that is being exchanged or transferred and since these legs of the transactions can be subject to failures, there should have been some discussion on maintaining the data integrity in case of failure and possible recovery strategies for the same. However, the fact that the authors have considered the system in Toto, it can be argued that data loss is part of the system’s abilities to heal itself and recover from failures. CRITIQUE OF SECOND PAPER The paper to be critiqued is “Enhancing Web Services with Diagnostic Capabilities” by Dupre et al. SUMMARY OF THE PAPER The focus of this paper is on identifying the sub-services that can be used to diagnose the web services effectively in case of failure. The paper also focuses on the adoption of industry wide standards for web services that would introduce protocols and other standards that are adopted worldwide and would lead to common standards for each of the services being used by the participants in the industry. As the authors state in the introduction to the paper, “we discuss a protocol for a global diagnoser service, and we characterize the operations that local diagnoser must support in order to comply with such a protocol. The goal is the identification of the faulty service, not debugging the service itself. In addition, the local diagnoser may identify a part of the service which is claimed to be responsible for the fault. An early fault detection and a fault identification which is as precise as possible are a necessary precondition for a better recovery from faults” (Dupre et al, 2005). PROBLEM ADDRESSED AND RESEARCH CHALLENGES As with the previous paper, this paper takes the example of an online book store for the purposes of illustration and to provide justification for the web service architecture to diagnose the faults and take appropriate action. The research problem being discussed is the ability of the sub-services to respond to faults in the system and diagnose the same. Further, the issue of diagnostics with its set of complex interactions between the dependent variables that make up the dynamic model of the system is also discussed. The research challenge lies in the ability of the proposed architecture to tackle the issue of diagnosis of fault based issues that arise out of the complex web services that support the systems. The system complexity is a given and the faults that occur in the transactions that are processed by the system are to be diagnosed using advanced models that rely on algorithms to do the job. MAIN CONTRIBUTION OF THE PAPER The main contribution of the paper is the way in which the paper tries to address the issue of diagnostics of faults and does not assume that the architecture would be able to diagnose all problems effectively. As the paper states, “Under worst-case complexity analysis, the problem of finding all diagnoses, or even a single diagnosis, is intractable. In practice, the actual occurrence of combinatorial explosion heavily depends on the model and on the availability of discriminating observations; it is definitely more likely to occur for reasoning on dynamic models. It can be avoided or mitigated in several ways” (Dupre et al. 2005). Hence, the emphasis is on finding the root causes based on dynamic models that recognize the need for analysis of complex situations by resorting to an architecture that assesses the impact of local diagnoser as well as global diagnoser. It is this combination of the local and the global that is appealing to the readers who want to know how diagnosis of faults can take place in a system that is complex. STRONG ASPECTS OF THE PAPER The strength of the paper lies in the ability of the authors to delve deep into the technological aspects of the research problem and come up with solutions that are based on a sound theoretical underpinning as well as on robust technical details that cover all aspects of the research problem at hand. The solutions suggested by the authors are comprehensive and lead to further study in the field of analysis of complex systems and the interactions in the architecture of the same. The other strengths of the paper are the analysis of input and output parameters that interact with the underlying structure of the system in producing a unified picture of the complex systems and their internal interactions as well as external interfaces. The key strength of the paper is its ability to state the research problem articulately and address the same by means of parameterization of the component based model as a diagnostic tool. WEAKNESSES AND SUGGESTIONS FOR IMPROVEMENTS Though the above section pointed out the technical depth of the paper, the fact that the paper could have been a bit less jargon based could have gone some way in making the paper more readable. The point here is that technical analysis is by itself a good feature but if it ends up confusing the readers, the purpose of depth is lost. As the previous section made mention of the excellent algorithms that were used to explain the concepts, the way in which the written explanations accompanying the algorithms were used could have been made in simpler language. SURVEY OF TWO PAPERS This section attempts to provide a survey of the two papers that were critiqued and come up with a description of the research problems and the challenges/ ABSTRACT The topic at hand is the issue of web service architectures and their reliability for conducting e-commerce transactions. The focus of both the papers has been on the way in which the systems recover from failures and the diagnostic capabilities of these systems in managing the recovery of the systems after failure. While the first paper has focussed on the ability of the systems to heal them after failure, the second paper focuses on the diagnostic capabilities of the systems in identifying faults and treating them appropriately. INTRODUCTION It is the case with many e-commerce applications that transactions fail before all the legs of the transaction are completed. This leads to situations where it becomes difficult to restore the system to the original state or repair the loss arising out of the failure of the transactions. Hence, it becomes imperative to have fool proof strategies for failure and have recovery capabilities. The two papers address these issues from different angles. While the first paper focuses on the ability of systems to self heal after failure, the second paper addresses the issue of diagnostic capabilities of the systems in identifying fault lines and taking appropriate decisions. Both the papers essentially provide solutions to the issue of transaction recovery after failure. While the first paper discusses the ways to adapt the web services to recover from system failure, the second paper talks about the component based model that can be used to diagnose the faulty transactions and provide for recovery of the same. MAJOR RESEARCH CHALLENGES IN THE AREA The major research challenge in the area of transaction recovery is the ability of the systems to heal them as is discussed in the first paper and the way in which the systems can be diagnosed for root causes as is the case with the second paper. Hence, the issues of self healing systems and diagnosis of faults and the subsequent attempts to recover the system from crashes and failure represent the key challenges in this area. There is no doubt that further research into these areas is a necessary step towards achieving completeness in terms of solutions (proposed and actual) and that the future research challenges would arise from the way in which systems interact with each other and internally and externally and hence, solutions to these must concentrate on a holistic approach to the same. CRITIQUE OF EXISTING WORK The existing literature on system recovery after failure concentrates on data backup and maintaining of data integrity when the systems fail. In many of the proposed solutions, the emphasis is on data storage and the relational models of databases that are used to predict the recovery of systems from failure. While the existing work on these areas cannot be deemed incomplete, the papers that were critiqued are definitely a step towards addressing some of the gaps in the present research. So far, the existing work has tended to focus on the consistency of the transactions and the reliability of the transactions by providing roll back options when the transactions fail. These two papers focus on recovery as well as rollback and hence are a novel approach towards the situation at hand. The emphasis on recovery is as important as the emphasis on roll back and particularly when the transaction costs are involved, recovery is a much better strategy as compared to roll back. CONCLUSION The papers that have been critiqued here represent important additions to the existing literature on transaction recovery and system diagnosis in case of failure. In times when system failure and the ways in which the systems recover from failure occupy the mind space of CIO’s around the world, these papers represent attempts to provide solutions to these issues. The papers are valuable additions to the existing literature on system recovery and the solutions presented in the paper are certainly worth considering. The issues here have a lot of impact on the ability of the systems to provide fail proof services to the users. Hence, it is imperative that action on these fronts must be taken up by the CIO’s and the software departments in the areas of web services to provide effective and lasting solutions to the issues being discussed. In conclusion, the papers address two dimensions of the same issue and are well formulated and the solutions articulated in a logical and cogent manner. Sources D. Ardagna, M. Trubian, and L. Zhang. SLA Based Profit Optimization in Multitier Systems. In NCA 2005 Proc., 2005. L. Ardissono, L. Console, A. Goy, G. Petrone, C. Picardi, M. Segnan, and D. T. Dupre´ . Advanced Fault Analysis in Web Service Composition. In International WWW Conference, poster session, pages 1090–1091, 2005. L. Ardissono, L. Console, A. Goy, G. Petrone, C. Picardi, M. Segnan, and D. T.Dupre´ . Enhancing Web Services with Diagnostic Capabilities. In ECOWS05 Proc., 2005. D. Bianchini, V. De Antonellis, B. Pernici, and P. Plebani. Ontology-based methodology For e-service discovery. Accepted for publication on Journal of Information Systems, Special Issue on Semantic Web and Web Services, 2004. Y. Pencol´e and M. Cordier. A formal framework for the decentralized diagnosis of large Scale discrete event systems and its application to telecommunication networks. Artificial Intelligence, 164(1-2), 2005. R. Reiter. A theory of diagnosis from first principles. Artificial Intelligence, 32(1):57–96, 1987. N. Roos, A. ten Teije, and C. Witteveen. A protocol for multi-agent diagnosis with Spatially distributed knowledge. In 2nd Int. Joint Conference on Autonomous Agents and Multi-Agent Systems (AAMAS-2003), Melbourne, Australia, July 2003. W. van der Aalst and K. van Hee. Workflow Management - Models, Methods, and Systems. The MIT Press, 2002 Read More