Are you running #PgBouncer? Are you disappointed that it is slow + unscalable? @butro's latest VLDB paper shows how to use #eBPF to accelerate this popular PostgreSQL proxy by running on kernel threads.
➡️Paper:
➡️Code: vldb.org/pvldb/vol16/p3…
github.com/mbutrovich/tig…
.@butro's PhD thesis is on user-bypass methods to embed DBMS logic into the kernel with eBPF. PgBouncer does auth but then connection pooling + forwarding happens in kernel. Tigger removes copying data from kernel to app. It also gets multi-threading for free using kthreads.

Vaccination DB Talk Booster #12: @majetideepak lays out the architecture of Velox:

Velox is a C++ vectorized execution engine to replace @prestodb's Java-based engine. It is being developed by @AhanaIO + Facebook + others. Velox is not a full DBMS: it doesn't have a SQL parser, catalog, optimizer, networking, etc. It's an extensible execution engine + buffer pool + storage connectors. It internally uses @ApacheArrow columnar format, but not the rest of the Arrow engine.

If you're interested in DB internals, stop what you're doing and watch @CMUDB Quarantine Talk from Nico+Cesar about @SQLServer's Cascades query optimizer:

Many talks this semester were good. This one is the best. My thread provides key takeaways [6:50] Microsoft hired Goetz Graffe in the 1990s to help them rewrite original Sybase optimizer into Cascades. This framework is now used across all MSFT DB products (@SQLServer, @cosmosdb, @Azure_Synapse).

So I was slightly involved with this story:

https://twitter.com/Werner/status/1054901529478459392

TLDR: I've read the internal Amazon document in the CNBC story. @Werner is correct. When I told the reporter that it wasn't a big deal he called me on the phone and flipped out. Two weeks ago this reporter reached out to the top DB profs (@joe_hellerstein, @franklinmj, @pateljm) + junior people like me asking for help with a story about Amazon Aurora. His original email said that he was looking for helping on understanding database terms.