Oracle Analytics and AI

Chad Williams

Hello,

We've been looking at possible moving from the traditional RPD to the Semantic Modeler for our data modeling practice. Got everything set up with Git and it looks great.

What I'm wondering is… are there any best practices or even approaches that anyone has come up with for working with your semantic model across a full deployment lifecycle environment (with Dev, Test, QA, and Prod)? Specifically trying to figure out how we would do our development work in Dev (for example) and then deploy those changes (or model?) upstream to Test, QA and finally Prod.

Would you maintain a separate repo for each environment, or just pick one to work from for development, and then move that model up using the the deploy feature to move into higher environments? Are there are any considerations here that one needs to watch for (ie I think database connections could be tricky).

Thanks in advance,

Chad

Gianni Ceresa

I would follow the direction hinted by @Ambient .

There isn't best practices for the semantic model explicitly, because the idea of the native git integration is to be able to use classic git strategies and best practices decided in your organisation.

If somebody else there already use git for some other product, you could go have a chat with them to see how they use it.

If you look at the various strategies listed above, they often lookalike, but each has some benefits in one thing or the other.

I wouldn't work on separate repositories for your 4 environments, because you kind of lose the benefits of git. You could use branches for your environments if needed, or tags as well to identify what make a release of all the set of features you then want to push to prod etc.