We are coming up with mundane processes of replicating transforms, datasets, workshops, etc. in order to properly achieve UAT without impacting existing production dataflows. There has to be a better way with Ontology branching or something. Will Palantir please provide a best practice process they hope customers/users will use as a simulated ‘dev’ / ‘test’ environment?
You can currently use ontology branching to simulate changes on OMA. We’re currently working on a more unified approach to branching across apps, that should allow for more complex workflows to be successfully branched and trialed across Foundry.
I look forward to seeing this update and hope that it comes with a solid best practices documentation or video as well.
Another way to do this is create different spaces and corresponding Ontologies. You can treat your current space and Ontology as the ‘dev’ environment, a new one as the test, and so on…
You can then use Marketplace to package and deploy all assets (pipelines, workshops, ontology, etc.) over from dev to other environments as “releases”. You can also lock the test/prod workshops so that they can only be edited through dev.
That’s exactly what I was thinking of doing in harnessing marketplace to package and deploy from one to the other, but with the emergence of branching and more I was curious what Palantir was working on and what they believe the best approach today and hopefully soon after with their more unified solution. I think what you said though is the best way to do it with what’s available in platform right now, so that’s likely better than the approach some of my colleagues are trying to do.