How to Test New Software on a Data Lake at Netflix Scale

Lake FS saved us from the analysis paralysis of overthinking how to test new software on our data lake at Netflix scale. In less than 20 minutes, I had Lake FS up and running was able to run tests against my production data in isolation and validate the software change thoroughly before pushing the production. It's because it's open source. Yeah. People are talking about testing a lot. So you have like these issues with testing where you can't test on like, if you implement an algorithm and then you have like results from that algorithm in the training data going forward is messy quickly. And so being able to roll back would be nice.