This is the page to describe how to integrate Medici 2 and DataWolf with supporting IRODS. Through this integration/refactoring, we will achieve
- DataWolf will use Medici 2 as Dataset storage.
- But, the design/refactoring will allow for DataWolf to store workflow in Medici 2
- Medici 2 will use IRODS as file storage
- Medici 2 currently has two implementation for file storage: MongoDB GridFS and local file system
- It is possible for us to store the part of the meatadata of datasets/files to IRODS, too. But it is a low priority
Implementing IROD data storage in Medici 2
Medici 2 has the file storage API. It has two implementations: local file system and mongoDB GridFS. We need to have implementation for IRODS. Current DataWolf has the IRODS implementation for file storage by using IRODS java client library, jargon, already. We could utilize the DataWolf implementation
Refactoring DataWolf to use Medici 2 as dataset storage
In order for DataWolf to use Medici 2 as dataset stroage, the following tasks needs to be done
Researching on Spring Data vs Guice
Regarding eclipse (OSGI) support and properties configuration support.
Separating the service layer and implementations
- Remove current JPA annotation from beans (separating spring data)
- workflow, workflow step
- workflowtool
- dataset
- file
- Defining service API as abstract supporting (DAO -> service and implementation)
- CRUD and extra capability (save, load, remove, update etc)
- Implementation layer for each service
- JPA (without annotation in the bean but with xml definition)
- Mongo
- Medici 2