Arquiteturas distribuídas para modelagem e simulação podem escalar a execução de modelos grandes e complexos. Essas arquiteturas freqüentemente utilizam estratégias de ponto de verificação para garantir a execução de componentes síncronos e assíncronos. No entanto, a evitação completa de pontos de verificação inúteis é impraticável e pode reduzir drasticamente o desempenho da simulação. Neste artigo, apresentamos um conjunto de métricas para identificar pontos de verificação inúteis em tempo de execução. Além disso, estendemos uma decisão probabilística que emprega nossas métricas propostas para criar apenas pontos de verificação com alta probabilidade de serem carregados por operações de reversão. O método identifica pontos de verificação inconsistentes com base nos padrões de comunicação e na granularidade dos eventos desde a última reversão. Os resultados mostraram que as métricas propostas permitem reduzir o número de pontos de verificação inúteis sem impactos negativos no desempenho da simulação e supera as estratégias probabilísticas tradicionais em termos de tempo de reversão.
Intent-Based Networking (IBN) is showing significant improvements in network management, especially by reducing the complexity by using intent-level languages. However, IBN is not yet integrated and widely deployed in most of the large-scale networks. Network operators may still encounter several issues deploying new intents, such as reasoning about complex configurations to understand previously deployed rules before writing an intent to update the network state. Large-scale networks may include several devices distributed in multiple hierarchical levels of its topology; each of these devices is configured using low-level, vendor-specific languages. Thus, inferring intents from these low-level configuration files can be an arduous and time-consuming task. Current solutions that derive high-level representations from bottom-up configuration analysis can not represent configurations in an intent-level. Moreover, they fail by not aggregating essential details to enhance the representation. In this work, we present a bottom-up process to extract intents from network configurations. To validate our approach, we develop a system called SCRIBE (SeCuRity Intent-Based Extractor), which decompiles security configurations from different network devices and translates them to an intent-level language called Nile. To demonstrate the feasibility of SCRIBE, we outline two case studies and evaluate our approach with dumps of real-world firewall configurations containing rules from various servers and institutions. Results show that our solution can represent configurations in intent-level and also maintain a high accuracy representing aspects of low-level configurations.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.