Xpress Insight 5.10
Xpress Insight 5is the technology component of Xpress Optimization that allows you to rapidly configure business applications to operationalize your analytic and optimization models. It provides a Compute capability that enables remote execution of Xpress Solver and Xpress Mosel. It is built on an industry-standard architecture that is scalable, secure and resilient.
This release focuses on execution performance enhancements by providing lightweight execution modes, introducing resource groups, allowing for app promotion on FICO Platform, and finally by providing Insight docker images for local development.
Xpress Insight 5is the technology component of Xpress Optimization that allows you to rapidly configure business applications to operationalize your analytic and optimization models. It provides a Compute capability that enables remote execution of Xpress Solver and Xpress Mosel. It is built on an industry-standard architecture that is scalable, secure and resilient.
This release focuses on execution performance enhancements by providing lightweight execution modes, introducing resource groups, allowing for app promotion on FICO Platform, and finally by providing Insight docker images for local development.
- Lightweight model execution modes allow the model developer to define which parts of the data model need to be populated at the start and which need to be captured at the end of a scenario execution. This ensures only the necessary data is transferred, which improves overall performance.
- Caching of model-resources folder: In cases where the model resources folder stores a lot of data, them being cached on a worker reduces waiting time of users for a scenario execution to complete. Furthermore, it helps to improve worker performance and resilience.
- New concept of resource groupes gives Insight administrators granular control over thread-based scheduling.
- All execution modes are grouped into resource groups. For these resource groups, the thread consumption of all their execution modes can be configured.
- The thread annotation on execution modes is deprecated.
- A model can query the values programmatically.
- For existing and new apps, the creation of resource groups and assignment of execution modes is automated.
- For a single Insight app installed on different sites, the administrator of each site can now configure not only the worker’s thread capacity but also the threads per resource group.
- All execution modes are grouped into resource groups. For these resource groups, the thread consumption of all their execution modes can be configured.
- Memory aware scheduling as part of resource groups:
- Besides threads, Insight scheduler can optionally honor “memory”.
- A resource group can optionally declare the memory requirement for its execution modes, while a worker can declare a memory capacity.
- Note: this is an honoring model. The model developer needs to make sure that not more memory (or threads) are used than a resource group has declared.
- Besides threads, Insight scheduler can optionally honor “memory”.
- (FICO Platform only): Granular control of app promotion for Xpress Insight lifecycle stages.
- The Xpress Insight Launch Page in FICO Platform shows all apps per lifecycle stage and allows selection for which apps are to be promoted from one lifecycle environment to the next.
- This provides a granular control in a user-friendly interface for app promotion.
- Ready-to-use Docker images and docker files ease deployment for local development and testing:
- On Docker Hub, images for Xpress Insight are made available which allow you to run Xpress Insight via docker - this makes Insight available for macOS users besides Windows and Linux.
- Our new GitHub page provides a docker compose script which shows an example how to leverage these images locally.
- These are intended for local development for now and do not yet fully support company-wide installations.
Updated 2 years ago
Version 3.0