That was a good video. It could be more helpful for prospective data engineers if you could emphasize that Lambda/Kappa is not a direct replacement of traditional batch processing but rather extensions or evolutions. It could also be helpful to go over what kind of data sources can be still batch processed and what kind of data sources need near-real-time processing may be in the use case section.
How does Kappa architecture handle training heavy ML algorithms in the real time engine? Or it has to have a trained model available? Where and how shall the modeled be trained and deployed?
Thank you helped me understand difference in a quick way. I have a question in which scenario do we go for both real time and batch where the result is stored in same persistent database( and table).