March 18 2022 - DuckDB with Hannes Muleisen

sedaily https://softwareengineeringdaily.com/2022/03/18/duckdb-with-hannes-muleisen/ transcription http://softwareengineeringdaily.com/wp-content/uploads/2022/03/SED1439-DuckDB-with-Hannes-Muhleisen.pdf

even if we are running our of memory we are still able to complete de query successfully

“never give up, never surrender project”

gracefully degrade as we start using disk

traditional systems they will run happily in memory until they realize. oh this doesnt fit in memory anymore and then they switch to and entirely different algorithm for example: go from hash join to a sort merge join.

we don’t wanna have this hard cuts in out implementation because it creates this unexpected performance clips

people dislike about relational systems. query is the same but you add one more row of that and suddenly an entirely different algorithm is executed and they have no kind of handle on this.

correlated subquery

January 31 2022 - Scaling PlanetScale with Sugu Sougoumarane

https://softwareengineeringdaily.com/2022/01/31/scaling-planetscale-with-sugu-sougoumarane/

http://softwareengineeringdaily.com/wp-content/uploads/2022/02/SED1413-Scaling-PlanetScale-with-Sugu-Sougoumarane.pdf

Consensus protocol Flex Paxos

paxos is actually two algorithms working together