To identify the cause of the error and maintain the health of system, an administrator usually analyzes event log data since it contains useful information to infer the cause of the error. However, because today’s systems are huge and complex, it is almost impossible for administrators to manually analyze event log files to identify the cause of an error. In particular, as OpenStack, which is being widely used as cloud management system, operates with various service modules being linked to multiple servers, it is hard to access each node and analyze event log messages for each service module in the case of an error. For this, in this paper, we propose a novel message-based log analysis method that enables the administrator to find the cause of an error quickly. Specifically, the proposed method 1) consolidates event log data generated from system level and application service level, 2) clusters the consolidated data based on messages, and 3) analyzes interrelations among message groups in order to promptly identify the cause of a system error. This study has great significance in the following three aspects. First, the root cause of the error can be identified by collecting event logs of both system level and application service level and analyzing interrelations among the logs. Second, administrators do not need to classify messages for training since unsupervised learning of event log messages is applied. Third, using Dynamic Time Warping, an algorithm for measuring similarity of dynamic patterns over time increases accuracy of analysis on patterns generated from distributed system in which time synchronization is not exactly consistent.

Several fields of science have demanded large-scale workflow support, which requires thousands of CPU cores or more. In order to support such large-scale scientific workflows, high capacity parallel systems such as supercomputers are widely used. In order to increase the utilization of these systems, most schedulers use backfilling policy: Small jobs are moved ahead to fill in holes in the schedule when large jobs do not delay. Since an estimate of the runtime is necessary for backfilling, most parallel systems use user’s estimated runtime. However, it is found to be extremely inaccurate because users overestimate their jobs. Therefore, in this paper, we propose a novel system for the runtime prediction based on workload-aware clustering with the goal of improving prediction performance. The proposed method for runtime prediction of parallel applications consists of three main phases. First, a feature selection based on factor analysis is performed to identify important input features. Then, it performs a clustering analysis of history data based on self-organizing map which is followed by hierarchical clustering for finding the clustering boundaries from the weight vectors. Finally, prediction models are constructed using support vector regression with the clustered workload data. Multiple prediction models for each clustered data pattern can reduce the error rate compared with a single model for the whole data pattern. In the experiments, we use workload logs on parallel systems (i.e., iPSC, LANL-CM5, SDSC-Par95, SDSC-Par96, and CTC-SP2) to evaluate the effectiveness of our approach. Comparing with other techniques, experimental results show that the proposed method improves the accuracy up to 69.08%.

본 논문에서는 재활용 필름을 사용한 친환경 그리드형 보강재와 사전에 충진재의 혼입을 통해 분산성을 향상시킨 섬유를 도로 포장층에 복합적으로 적용한 공법의 현장 적용성을 분석하였다. 그리드형 보강재를 구성하고 있는 필름은 폐기된 PE수지를 재활용하여 친환경성을 강조하였고, 아스콘 혼입 섬유는 아스콘 충진재의 사전 혼입을 통한 섬유의 분산성을 확보한 것이다. 복합 섬유를 혼입한 교량 포장층을 시공함에 있어 발생할 수 있는 제반 상황을 사전에 인지할 수 있도록, Mock Up 포장층에 대한 기본적인 성능 분석을 실시하였다. 분석 결과, 친환경 그리드 보강재와 분산성 섬유를 함유한 포장층에서 강도 및 응력, 처짐 저항성에서 모두 기능성이 향상된 것을 확인할 수 있었다. 그러므로, 섬유 실제 현장 적용시 친환경성 재료 도입과 더불어 안전성 측면에도 효율적인 공법이라고 판단된다.