Imagine a conference with:
👉no fluff talks
👉no sales
👉community you actually want to talk with
Open your eyes. It's #POUG2026. Last tickets with 30% off:
poug2026.konfeo.com
#POUG2026 tickets are live. Two options:
🐦 Early Bird — 50% or 30% off. The conference. Best price, period.
🔬 Regular + Workshop — full price, but includes an exclusive hands-on workshop. Full-ticket holders only.
You pick!
poug2026.konfeo.com/en/groups
AI is only as good as what it knows. For Oracle Database, that's been a problem. Built a library of 100+ practical guides to fix that — covering everything from PL/SQL to ORDS to migrations. Open source, MIT license. github.com/krisrice/oracl…
New project released: WayFlow!🚀
WayFlow is a powerful and intuitive Python library for building AI assistants. It simplifies development, supports reusability, and is great for fast prototyping and iteration.
Get started: github.com/oracle/wayflow#AI#AIAgents#OracleOSS
Do you know that at the #POUG2025, you can win a lifetime free POUG ticket? This will be the award in our 10th anniversary celebration contest!
So, by buying a ticket now with a 20% discount, you might be purchasing a ticket for life to our conferences:
poug2025.konfeo.com
Vergiss langweilige Konferenzen mit Verkaufsmaschen und null Mehrwert 🙄
Komm zu einem 100 % Tech-Event direkt hinter der Grenze - und werde Teil einer einzigartigen Community. #POUG2025 wartet im September auf dich! 🔥
poug.org/en/
@FranckPachot@mortenbraten@StewAshton@gwenshap@connor_mc_d Isn't it about using the right tool for a job ? There are cases where OOP is the right paradigm, but if it is blindly used, it can cause a lot of overhead as well. ORMs are just an interface for OOP, and it is still better to use existing ORM than build database layer yourself
@mortenbraten@StewAshton@gwenshap@connor_mc_d So because the RDBMS can return only one two-dimension table per SQL query, the application must abandon its structured object model and flatten everything like in a text file, a single-table screen, or a cobol data division of the 80's? ORMs fix the limitations of the SQL API
Where do OLTP workloads spend most of their time?
In 2008 it was waiting for the CPU.
In 2024 it is waiting for the network.
Applications running on leading databases spend 23%-68% of their time on network communication!
(At least according to "OLTP Through the Looking Glass 16 Years Later" paper from CIDR 25)
Using DPDK (kernel bypass networking) significantly reduces overhead, but is challenging to implement. Using stored procedures reduces, but does not eliminate, the overhead.
The paper also explores the biggest issue with stored procedures - the lack of isolation. They explore several isolation models and show how introducing more isolation also increases the communication overhead.
The only problem with the paper: It is focused on VoltDB, which no one uses, and extrapolates to more popular DBs. Would be interesting to see a PG-centric version of this.
Jeśli ktoś nie załapał się wykład Dragon Sectora o pociągach na żywo część 2, to tu można obejrzeć nagranie, także z tym slajdem, który już przeszedł do legendy.
streaming.media.ccc.de/38c3/relive/336
Gorąco polecamy za wartości poznawcze i komediowe.
#deliberatorium wiara ze edukacja zadziala, to troche zyczeniowe myslenie. Obie strony musza byc w mediach spolecznosciowych bo tam odbywa sie walka. Blokady nic nie dadza bo za latwo je obejsc.
The computing behind LHC data processing (WLCG) these days:
💻 1.3M cores crunching data
📀 1 exabyte on disk
📼2 exabyte on tape
🌍 Distributed across 40+ countries
#BigData fueling collaboration and discovery. #CERN#LHC