Automated Software Engineering Group

Context: Model-driven engineering (MDE) promotes the active use of models in all phases of software development. Even though models are at a high level of abstraction, large or complex systems still require building monolithic models that prove to be too big for their processing by existing tools, and too difficult to comprehend by users. While modularization techniques are well-known in programming languages, they are not the norm in MDE. Objective: Our goal is to ease the modularization of models to allow their efficient processing by tools and facilitate their management by users. Method: We propose five patterns that can be used to extend a modelling language with services related to modularization and scalability. Specifically, the patterns allow defining model fragmentation strategies, scoping and visibility rules, model indexing services, and scoped constraints. Once the patterns have been applied to the meta-model of a modelling language, we synthesize a customized modelling environment enriched with the defined services, which become applicable to both existing monolithic legacy models and new models. Results: Our proposal is supported by a tool called EMF-Splitter, combined with the Hawk model indexer. Our experiments show that this tool improves the validation performance of large models. Moreover, the analysis of 224 meta-models from OMG standards, and a public repository with more than 300 meta-models, demonstrates the applicability of our patterns in practice. Conclusions: Modularity mechanisms typically employed in programming IDEs can be successfully transferred to MDE, leading to more scalable and structured domain-specific modelling languages and environments.

Comparison of large models can be time-consuming since every element has to be visited, matched, and compared with its respective element in other models. This can result in bottlenecks in collaborative modelling environments, where identifying differences between two versions of a model is desirable. Reducing the comparison process to only the elements that have been modified since a previous known state (e.g., previous version) could significantly reduce the time required for large model comparison. This paper presents how change-based persistence can be used to localise the comparison of models so that only elements affected by recent changes are compared and to substantially reduce comparison and differencing time (up to 90% in some experiments) compared to state-based model comparison.

The growing size of software models poses significant scalability challenges. Amongst these challenges is the execution time of queries and transformations. In many cases, model management programs are (or can be) expressed as chains and combinations of core fundamental operations. Most of these operations are pure functions, making them amenable to parallelisation, lazy evaluation and short-circuiting. In this paper we show how all three of these optimisations can be combined in the context of Epsilon: an OCL-inspired family of model management languages. We compare our solutions with both interpreted and compiled OCL as well as hand-written Java code. Our experiments show a significant improvement in the performance of queries, especially on large models.

Flexible or bottom-up model-driven engineering (MDE) is an emerging approach to domain and systems modelling. Domain experts, who have detailed domain knowledge, typically lack the technical expertise to transfer this knowledge using traditional MDE tools. Flexible MDE approaches tackle this challenge by promoting the use of simple drawing tools to increase the involvement of domain experts in the language definition process. In such approaches, no metamodel is created upfront, but instead the process starts with the definition of example models that will be used to infer the metamodel. Pre-defined metamodels created by MDE experts may miss important concepts of the domain and thus restrict their expressiveness. However, the lack of a metamodel, that encodes the semantics of conforming models has some drawbacks, among others that of having models with elements that are unintentionally left untyped. In this paper, we propose the use of classification algorithms to help with the inference of such untyped elements. We evaluate the proposed approach in a number of random generated example models from various domains. The correct type prediction varies from 23 to 100% depending on the domain, the proportion of elements that were left untyped and the prediction algorithm used.

A comparison of four protein model-building pipelines (ARP/wARP, Buccaneer, PHENIX AutoBuild and SHELXE) was performed using data sets from 202 experimentally phased cases, both with the data as observed and truncated to simulate lower resolutions. All pipelines were run using default parameters. Additionally, an ARP/wARP run was completed using models from Buccaneer. All pipelines achieved nearly complete protein structures and low Rwork/Rfree at resolutions between 1.2 and 1.9 Å, with PHENIX AutoBuild and ARP/wARP producing slightly lower R factors. At lower resolutions, Buccaneer leads to significantly more complete models.

A key challenge in porting enterprise software systems to the cloud is the migration of their database. Choosing a cloud provider and service option (e.g., a database-as-a-service or a manually configured set of virtual machines) typically requires the estimation of the cost and migration duration for each considered option. Many organisations also require this information for budgeting and planning purposes. Existing cloud migration research focuses on the software components, and therefore does not address this need. We introduce a two-stage approach which accurately estimates the migration cost, migration duration and cloud running costs of relational databases. The first stage of our approach obtains workload and structure models of the database to be migrated from database logs and the database schema. The second stage performs a discrete-event simulation using these models to obtain the cost and duration estimates. We implemented software tools that automate both stages of our approach. An extensive evaluation compares the estimates from our approach against results from real-world cloud database migrations.