A Type-Sensitive Service Identification Approach for Legacy-to-SOA Migration

Fenrir: Blockchain-Based Inter-Company App-Store for the Automotive Industry

“…First, the complete transformation of automotive software is challenging to achieve in the short term, so we must study how to integrate legacy software into the proposed new deployment pipelines during the transition stage. A promising approach would be automatic parsing and wrapping mechanisms allowing for adapting and standardizing legacy packages, such as in [52]. Second, developer-declarative inter-software dependency potentially poses a threat when increasing the number of authorized services; thus, Fenrir will require some statistical model allowing potential failures that might be unnoticed in the development phase to be traced and detected [33].…”

From legacy to microservices: A type‐based approach for microservices identification using machine learning and semantic analysis

“…• Metrics and relations: Our first initiative was to use the same class-level metrics and method-level metrics M, which were used in a rule-based service-type detection approach, 32 to obtain the feature matrix X jVjÂjMj . The method-level metrics were used to compute some class-level metrics.…”

“…For example, Abdellatif et al 23 proposed a type‐based service identification approach that extracts Utility, Entity, and Application services from legacy software systems. Through static analyses of the source code, they extract several types of interclasses relationships to build the call graphs of the systems and generate clusters that are considered as potential services.…”

“…They filter and classify each cluster using code metrics and a hierarchical detection rules‐based system to identify services and their types. We extend the work of Abdellatif et al 23 and use the typed services as a basis to identify the microservices inside monolithic systems. Different from the approach of Abdellatif et al, we rely on an ML classifier and a clustering algorithm to identify the typed services that will be then aggregated to form the final microservices.…”

“…We generate the feature matrix by one of the following methods: Metrics and relations: Our first initiative was to use the same class‐level metrics and method‐level metrics , which were used in a rule‐based service‐type detection approach, 32 to obtain the feature matrix . The method‐level metrics were used to compute some class‐level metrics.…”