Difference between revisions of "Auto IWG WP2"
|Line 6:||Line 6:|
== Intent ==
== Intent ==
engineering models (e.g., AUTOSAR) tend to become so big that they are hardly manageable with existing Eclipse technology. The results of this work package are meant to lay the basis for significantly improving the performance characteristics of Eclipse-based tools when handling automotive engineering models of today and making them proof for the future where model sizes are expected to grow up even more.
== Concept ==
== Concept ==
Revision as of 20:51, 31 January 2012
WP2: Big Model Support
This is work package 2 of the Automotive Industry Working Group.
- WP Lead: ITEMIS (S. Eberle)
Real world automotive engineering models (e.g., AUTOSAR) tend to become so big that they are hardly manageable with existing Eclipse technology. The results of this work package are meant to lay the basis for significantly improving the performance characteristics of Eclipse-based tools when handling automotive engineering models as of today and making them proof for the future where model sizes are expected to grow up even more.
Improvement approach 1: Integration of CDO in Sphinx Workspace Management
Sphinx is a general purpose integrated modeling tool platform. It is the basis for the Artop project which provides a platform for AUTOSAR design tools. Currently Sphinx (and Artop) are designed to work on models that are persisted in files. However, when the models instances become big, i.e, several tens or hundreds of mega bytes, it becomes very hard to handle them in an efficient way. Long loading latencies and excessive memory consumption are the almost unavoidable consequences and give users a rather poor impression of the tool or even block them from accomplishing their work.
The idea behind this improvement approach is to significantly improve the situation by leveraging the model repository technology CDO. CDO enables models to be persisted in locally or remotely located databases and provides complete support for model version management. A key enabler for handling big model instances are CDO's advanced memory management capabilities. They ensure that only the interesting part of a model gets loaded, support on-demand loading of other models parts later on, and guarantee that model parts that are no longer needed can be unloaded and garbage collected. The proposed approach for addressing the previously mentioned performance problems is to setup and use a local CDO model repository that is used as platform-internal model cache by the Sphinx Workspace Management. From the users' point of view the models will still appear file-based, but Sphinx actually imports/exports them to/from the CDO-based model cache and Sphinx-based tools use the latter for performing all editing and processing operations. This way, Sphinx-based tools automatically benefit from CDO's memory management capacities and come up with a much better performance when operating on big models.