Installation process is described here: https://help.hybris.com/6.4.0/hcd/478f702a00104636b16d2614d1db30ac.html. This experts forum topic may clarify your futher questions https://answers.sap.com/questions/12762979/datahub-where-to-place-opt-folder.html?childToView=127944...
As to the flow, it's a lot to explain. I would recommend to start from here https://help.hybris.com/6.4.0/hcd/8ba79fcc86691014a83e8530484d3892.html and then ask specific questions when you have trouble.
Overall the process is like this: you model your raw item model after the data structures existing in the source system, from which they're loaded into DataHub. Model target items for the data structures in the destination system, to which DataHub will publish data. Canonical model is sitting in the middle between raw and target. Usually it's highly normalized internal storage of data inside DataHub. It's the core of all your models and you actually have to do it first before raw and target modeling. The problem to come up with your canonical model is that it's neither raw nor target but rather something in the middle, so you may have difficulty to come up with items and their attributes in that model. Try to make is similar to the target model first. Then you will map your raw items to canonical items and define necessary transformation. Same is true for the target model - you will then map target items and their attributes to the canonical items and their attributes. Then you can adjust the canonical model and the transformation mappings. For example, you may have values in target model, which should be translated from the values in the canonical model, - the solution might be to create a lookup table (item) in the canonical model and use resolve() expressions for the target item transformations. Details are in the document, it's too much to explain everything here. Last note, if the destination system is hybris ERP, then when modeling target items think about what kind of impex you need to load the data by an impex file because that's what essentially is happening.
While testing and debugging your models, try to use as small data set as possible because understanding what happened with your items is very difficult. Load raw items - check RawItem table (or ReST API) to make sure they saved as you expect. Then run composition and verify that CanonicalItem contains expected data. Then publish and see if ERP has the data you expect. Do debug impex generated during publication you may use test-adapter extension (search for it in the doc). In the process you will tweak the transformation expressions and possibly the item structures. Hope this helps.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.