Tanel Poder 🇺🇦

I've been busy in the last months! Announcing a brand new "Practical Exadata Performance" training course! I'm not aware of any other training *like this* offered in the world. 😎 Also, a *free* seminar! Exadata is an awesome platform, if used right! learn.tanelpoder.com

The LONG data type in Oracle Database is tricky to work with Sadly the partition high value column uses LONG 26ai added CLOB and JSON version of it But what if you're on older releases? @TanelPoder shows how to convert LONG to VARCHAR2 with PL/SQL tanelpoder.com/posts/oracle-g…

bpftop 0.9.0 is out. It now displays bpf_link_info for attached programs, and there's a new installer script. github.com/jfernandez/bpf…

A 30-year SQL issue: normalization requires joins across the relational schema, and for performance, those joins should run in the RDBMS, not in the application. But in-database joins do not raise NO_DATA_FOUND or TOO_MANY_ROWS... linkedin.com/posts/franckpa…

@trollied @lemire Some engines (talking about Oracle's In-Memory Database module with SIMD) dynamically load the shared library for appropriate architecture on engine startup... tanelpoder.com/2014/10/05/ora…

@lemire @TanelPoder Why can’t compilers make binaries that initially detect capabilities at runtime, and follow whatever code path that is supported?

Almost all x64 processors today (Intel/AMD) support advanced instructions such as BMI2, AVX2. Unfortunately, most Linux distributions give you code that is compiled to run on systems without these instructions. It means that you may get inferior performance. Many distributions like RedHat have begun defaulting on processors that have these advanced instructions. Recent versions of the popular Linux distribution Ubuntu offers version of your favourite software compiled with these advanced instructions (x86_64-v3 packages). Unfortunately, as far as I can tell, they are disabled by default and require some work to be enabled. I invite you to experiment.

Just in the nick of time, it's my round-up of Interesting Links for April 2026! There's links about all sorts of stuff around Data Engineering, CDC, Iceberg, Data architectures, and even a little hint of AI (but only the useful kind, not the gross hype-y kind). rmoff.net/2026/04/30/int…

@thatjeffsmith There are a few little known features about SQL Monitoring (both Exadata specific and general), currently working on some new material & a blog entry will follow at some point too :-)

@thatjeffsmith Haha, when I wrote this, was also wondering if anyone reads it "pretty decent" vs "recent" (and pretty awesome source of info and references)

A pretty recent summary article from Oracle on enabling and getting SQL monitoring reports in various ways (including historical captures): Real-Time SQL Monitoring: a MUST for SQL Tuning blogs.oracle.com/coretec/oracle… (I'm working on something new & related too ;-)

Looking to make your DuckLake lakehouse fly? DuckLake 1.0 shipped earlier this month! DuckLake is an open lakehouse format brought to you by the creators of DuckDB. The format is built to be simple and fast, and a few thoughtful configuration choices can take it to the next level. Tomorrow at 9am PT, Alex Monahan is going live to cover getting started, plus the settings and strategies to get your lakehouse going fast. Live Q&A too. Register: luma.com/ducklake-1-0?u…

It's official, our Skills for @OracleDatabase github.com/oracle/skills Get started, simply: npx skills add oracle/skills

Update one table using a column from another with direct joins in 26ai: UPDATE <target_tab> SET <target_col> = <source_col> FROM <source_tab> WHERE <join_condition> You can also use these joins in DELETE statements

.@addyosmani shares hot takes: "A decent model with a great harness beats a great model with a bad harness" "The gap between what today’s models can do and what you see them doing is largely a harness gap." "A harness is a living system, not a config file you set up once" addyosmani.com/blog/agent-har…

@ashoKumar89 Horizontal scaling is nowhere near as reliable as vertical scaling. Running less machines, less processes, less disks, less network hops is always easier.

IPI stands for "inter-processor interrupt" here...

RHEL note relevant for Oracle workloads with hugepages + concurrent process startups/exits. IPI storms due to a specific kernel security patch. "Performance impact on hugepage workloads in RHEL 8 and RHEL 9 systems with high CPU counts" access.redhat.com/solutions/7132…

AI agent skills are a set of markdown instructions explaining how to do a task @krisrice has built a library of 100+ skills for working with Oracle AI Database with actionable examples, best practices, and common pitfalls Get them at github.com/krisrice/oracl…

One SQL command to spin up a petabyte-scale lakehouse. DuckLake 1.0 just shipped in @DuckDB, and it's live on MotherDuck today! DuckLake is built on a simple but profound idea: lakehouse metadata belongs in a SQL database, not thousands of JSON files. 1.0 brings multi-engine support, data clustering, fast streaming updates, and more 👇 motherduck.com/blog/announcin…

Ducks have 360-degree vision, which means we see every single one of you who voted to get MotherDuck to the GeekWire Awards finals. Thank you 🙏 Now we need one more push from the community to win🥇 Voting is open through April 16. If MotherDuck has made your data work easier, faster, or just a little more fun — now's the time to let us know. Vote here: geekwire.com/events/2026-ge…