Hello, We are evaluating Datahub and trying to es...
# getting-started
d
Hello, We are evaluating Datahub and trying to estimate what is required to implement and maintain it. We are now evaluating how to extend the metadata model, since our reality is a bit different from what the model presents this is a big issue for us. The reason is we are mainly focused in small data and work with in-house developed tools. So my questions refer to the extension of the model. 1) Step number 6 of the documentation requires that the GraphQL & React models are extended to contemplate what was incorporated to the model in the previous steps. The question is, how is this an optional step? Are the new entities available for querying and the user interface if not done? 2) This is more of a conceptual question... Why is it necessary to redifine in GraphQL & React what was already defined in Pegasus? We see this as an error-prone task which could (should?) be automatic. Is there something we are missing? If not, is this already identified and is it on the roadmap? 3) Similarly, we don't understand why a rebuild in necesary after the model is extended (step 5). We are also concerned that it unnecesarily affects the production system to incorporate new concepts (we see it analogous to restarting a DMBS after a table create). Again, are we missing something here? Thanks a lot. Regards, Agustín Mullin
g
Hello @damp-oxygen-37726! Welcome to the community. 1+2) The react & grapqhl steps are listed as optional because you still can take advantage of the REST apis without them. In the future we aim to remove the need to manually update graphql & react. In that world, we can either codegen or dynamically construct the logic to display new models in the UI. It is not on the roadmap because we are still figuring out how to sequence this work. Input from community members such as yourself is a key piece of that sequencing 🙂 What sort of models are you looking to bring to the UI? How critical are no-code models in the UI to the success of your Datahub deployment? 3) We also intend to add support for an API-based model creation system. In most cases, updating the metadata model is an infrequent occurrence. Since we find users are not frequently altering their metadata model, this has not been a priority.
d
Hello @green-football-43791, Thank you for your prompt and enlightening response. I agree that extending the model is not something which is done frequently, and so I understand it not being a top priority. Regarding how critical it is for our successful deplyment to have these models automatically generated, I would say it is not critical, but we see it as a potential problem. Regarding the models we would like to bring to the UI, we need to incorporate Business Intelligence and Reporting concepts/entities, with special attributes and relationships. We are just getting started with Datahub and maybe we have misunderstood something, but we have understood that the extension in step 6 is needed in order to see and edit these custom attributes in the UI, is that so? Thank you very much. Regards, Agustín
g
@damp-oxygen-37726 you are correct in your understanding- the step 6 extension is required in order to see or edit these attributes in the React UI
d
Thanks!