.@danavanaken's @VLDB2021 video on using OtterTune to automatically optimize an Oracle database in a production environment with @CBILIEN is now available. #VLDB2021 https://t.co/t3MKDdwbQF
OtterTune researchers are all over #VLDB2021 next week. Co-founders @andy_pavlo and @danavanaken are presenting papers, giving keynotes, award talks, and participating in panels on the future of ML for database tuning. See our site for the full schedule: https://t.co/E1jKxYHF2S
The #1 question that I've gotten from analysts + reporters in the lead up to our launch is how does @OtterTuneAI compare with Oracle's autonomous database offering. I discuss some of the differences / similarities in this interview with @nikosvg: https://t.co/Ts7BDDAXx0
This is @danavanaken's new #VLDB2021 paper on deploying OtterTune at @SocieteGenerale in collaboration with @CBILIEN. We used OtterTune on a production 1TB @Oracle DBMS and compare 3 ML algos.
TLDR: @OtterTuneAI outperforms human tuning but noisy cloud VMs made it tricky.
In certain SELECT queries, there were more rows returned than rows in the table, and in big numbers. Does that not really matter? Stopping jobs, creating temp tables, fake NULLs and so much more. Such egregious conduct. We will win a new contract!
We worked w/ @CBILIEN at a major French bank to deploy OtterTune on a 1TB @OracleDatabase v12 located on-prem. Real database, real workload, real enterprise hardware. The DBMS was also already tuned by expert DBAs.
Our paper with the results from this 10mo study is under review.
Before spinning out @OtterTuneAI as a startup we wanted to be sure it works in real-world. Prev research focused on older vers of @MySQL/@PostgreSQL. So @dvanaken teamed up w/ @CBILIEN to tune a prod @Oracle DBMS.
TLDR: OtterTune works. Sign up for a demo https://t.co/6TTHK8Rhc9
💣 Drop: After 6 years of research at CMU, I am spinning out the OtterTune project as a start-up with my students @danavanaken + @bohanzhang14: https://t.co/zT4RPOHSvw
Many thanks to our investors @HazyResearch@LuisvonAhn@MarcinZukowski@rxin @NikitaShamgunov and others!
We will also be posting new research results next month from our 10 month field study with @CBILIEN that shows how OtterTune successfully tunes a real-world DBMS deployment.
@FranckPachot You're lucky to have a Swiss driving licence. For a French that would have been a loss of 5 points. 12 lost = you loose your driving licence.