Zürich

I had a discussion about hashing with Petr Beles at the last Data Dreamland. He sad, for example, that DV Builder is prepared to calculate hash diff but there is no database, which they support for now, which would be quicker in hash computing than in column by column comparison. So, DV Builder currently doesn't calculate and store hash diff. But they also make a performance test with every new version of target database and they are prepared to switch the hash diff for selected db and versions. Interesting.

I must say, I don't remember, what data it was. But I know that it was in summer 1990, my first part-time job - development of an educational program in dBase III to demonstrate basic functions of this database. It was on PC - XT with 10MB HDD and 640 kB of memory. No. Now I remember. The most first data I worked with, were final grades for the whole secondary school, I've studied. I was a second year student, fond of computers and the school received a brand new computing laboratory of PCs. Friend of my class teacher prepared a database and a program in dBase III for final grades recording, mandatory reports and school reports printing. But there was no one who was able to use it, much less maintain it. So, in June 1990, last two weeks of school before holidays, I was at school but not in class. I acted as a team leader of group of ten schoolmates. We sat in the computing laboratory and copied all the grades from paper into database. Then we printed all the school reports. As I remember, it was about 1.200 students. It was crazy time. The first years after "velvet revolution" in former Czechoslovakia were very special.

I have, maybe, too simple point of view. If your data are more than 50% relational - go to Snowflake. Else - go to DataBricks. And don't tell me that you have 100% of data in semi-structured file extracts, although all your primary data sources are relational databases.

And what about DataBricks @Jonas De Keuster ?