Traudl

This looks pretty interesting. Like a mix of Figma/Miro with the video bubbles from pitch.com

Similar interesting is also the new Learned Index benchmark study: https://vldb.org/pvldb/vol14/p1-marcus.pdf

Throughput increased 54% for point lookups and 56% for scans

It looks like it found its way into Google Bigtable though: https://arxiv.org/abs/2012.12501

Yeah, apparently the results do not hold up under fair evaluation. They used a not well tuned RMI implementation. See: https://vldb.org/pvldb/vol14/p1-marcus.pdf

That said, PGMs are still interesting given their theoretical guarantees.

This could have huge implications for key/value stores and cloud storage systems as it provides a mechanism to access data by more than one key. The numbers to Amazon's z-order encoding also look very interesting. Of course, updates/inserts might still be a problem, but many of the other techniques (e.g, Amazon) are also static.