I’ve recently seen multiple, unrelated instances of people referencing Bf-trees. Good job, @benjdd.com.

Asking a coding agent to run `cargo build` and read referenced source files for context has made LLMs significantly more helpful and accurate at actually understanding why a compilation error is happening and being able to explain an appropriate fix. Much better than copy-pasting into online LLMs.

It’s frustrating how a bunch of database research from 1980s and before basically doesn’t exist anymore because it’s not on the internet, and it’s not even in the IEEE/ACM indexes of published work.

Does anyone have links to good writing on the sort of soft skills you learn from working in larger organizations about how to work in larger organizations as an IC? The overall space of soft skills dealing with the pretty common ways that large corporations behave.

Looking forward to reading about which disaggregated architecture HorizonDB aligned itself with

So what’s the feature set difference between pgDog, Neki, and multigres?

Our next event is on November 19th at StarTree’s office in downtown Mountain View. Come hear about Morel from Julian Hyde and and Query Optimization as a Service from Yuanyuan Tian! luma.com/xygolo9c

The recording from our last South Bay Systems meetup is now available! youtu.be/f1bz3efUJpM

@abigalekim.bsky.social @xiangpeng.systems and I are kicking off Madison Systems with a coffee chat on Sunday, Nov 9th. Come nerd out on systems! luma.com/v69tvpla

[PVLDB] Enhancing Transaction Processing through Indirection Skipping www.vldb.org/pvldb/v... Whereas VMCache improve pointer swizzing's complexity by removing the swizzling, this work points out that page and frame hints are highly effective, and okay if they're wrong.

This reminded me I've been sitting on draft blog posts about Copy-and-Patch JIT compilation for a while, and so I've finally published the first chunk of it: a minimal tutorial and explanation of how and why Copy-and-Patch actually works. Start at transactional.blog/copy-and-pat...

South Bay Systems returns on October 27th at Adobe in downtown San Jose. We have an Analytics-on-Object-Storage double feature this time starring two different Apache projects: Apache Pinot and Apache Doris. (Talk descriptions below.) Register now! luma.com/9o6bahgc

There was an accident with the recording where audio wasn't captured, so instead we can offer a recording from one of Jakob's practice runs on twitch: www.twitch.tv/videos/25845...

Had a fun time at the South Bay Systems meetup last night. Thanks @yugabytedb.bsky.social for hosting! @codedrift.social gave a great talk on WebAssembly: what it is (and isn't), how it connects to WASI, and promising projects. He cuts through a lot of the hype vs. reality. Recording coming soon.

[ASPLOS'25] Fusion: An Analytics Object Store Optimized for Query Pushdown www.cs.princeton.edu... Tightly integrating an Iceberg catalog with an object store means that one could make file-format aware erasure coding decisions, to permit pushing down filters and aggregations.

[VLDB] Towards Principled, Practical Document Database Design www.vldb.org/pvldb/v... If you've ever wished that there was a document database equivalent for relational databases' 3NF-style schema design guidance, then this is the paper for you.

[arXiv] On the Theoretical Limitations of Embedding-Based Retrieval arxiv.org/abs/2508.2... It's impossible to retrieve all combinations of pairs of documents post-embedding. Thus, there's usecases that vector search won't do well at. Conversely, BM25 excels in these cases.

I text-to-speech papers often, and www.paper2audio.com finally did the one thing that I was hoping AI would enable: replace tables/figures/diagrams with a summary of what is being shown. It makes table/diagram-heavy papers actually comprehensible. There's iOS and Android apps, and it's free.

[VLDB] NaviX: A Native Vector Index Design for Graph DBMSs With Robust Predicate-Agnostic Search Performance www.vldb.org/pvldb/v... It feels like a follow-on/improvement to ACORN. Also interesting to see HNSW built directly on a graph database working well.

Someone should go implement a bulk loading into btree mechanism relying on man7.org/linux/man-pa... to be able to prepare a tree of data, and then just atomically drop it into the main btree file as a sub-tree, as that'd be pretty cool to read about.

There’s surprisingly been no good citation for follower reads and the trade-offs therein. Super excited that this finally got published. law-theorem.com had “Coming soon!” for a few years 😭

For anyone else trying to catch up on DBSP, my recommended flow of learning is: 1. Watch the talk: www.youtube.com/watch?v=omOH... (h/t @wslim.bsky.social) 2. Read the spec/book: mihaibudiu.github.io/work/dbsp-sp... (h/t @avi.im) 3. Read the VLDB paper List is ordered by assumed knowledge of reader

In the Postgres-style MVCC vs MySQL-style MVCC debates, I'd really love to see an implementation of time-separated btrees (dl.acm.org/doi/pdf/10.1...) evaluated. It's CoW-BTree style "your path down the tree prunes out versions you don't want to see", but update-in-place and copies only on splits.

People often ask me about the differences in architecture between Amazon Dynamo (the 2007 SOSP paper), DynamoDB (the AWS serverless NoSQL database), and Aurora DSQL (the AWS serverless SQL databases). I memoized the response on my blog. brooker.co.za/blog/2025/08...

[arXiv] Theseus: A Distributed and Scalable GPU-Accelerated Query Processing Platform Optimized for Efficient Data Movement arxiv.org/pdf/2508.0... Great to see that Voltron Data folk writing about their GPU database!

Today’s NULL BITMAP is a very special one—I have been doing NULL BITMAP every week for two years, and to celebrate, this week I got a collection of friends to put together a printable zine of articles. I hope you enjoy it! buttondown.com/jaffray/arch...

To randomly sample a number of operations, one pulls from a PRNG. github.com/buildup-d... instead shows a cute trick for defining a stateless PRNG: pull RDTSC, run it through a quick hash to scramble the bits (e.g. rapidhash). Cache-miss-free, but you lose determinism in tests.

We had our biggest @southbaysystems.xyz meetup yet last night! Thanks to everyone who came, and thanks to Databricks for hosting! @andypavlo.bsky.social discussed the 50-year history of database tuning, applying AI/ML to the problem, and the future of auto-tuning (agentic reasoning, of course).

If you have a blog hosted on cloudflare pages and any part of your css looks missing only on safari, it's because of www.cloudflarestatus.com/incidents/ps..., and you have to go purge the cache to get rid of the cached wrongly-compressed asset files.

A good read on other, nicer ways that ISAs can represent vectorized loops: www.bitsnbites.eu/three-fundam...

I would be ever grateful if someone would be willing to go build a wrapper for PKGBUILD that is detached from any particular distro so that we can do away with Vcpkg/Conan/WrapDB/etc. and just re-use the same effort that distros put into packaging C/C++ dependencies.

I wrote up on bounding retries of write-write conflicts in MVCC+2PL. There is surprisingly little publicly written on how databases can retry, so doing my part to fill that gap in :) snowytrees.dev/blog/boundin...

Attention, South Bay folk! We have The Databaseologist, @andypavlo.bsky.social, giving a talk in the bay on August 6th. Come join us for a great time in hearing: ChatGPT Ain’t Got $%@& On Me! The Future of Automated Database Tuning Register now! https://lu.ma/ha0dc4nj

I had missed @ssougou.bsky.social's blog post series on consensus when it was originally posted. I really like the perspective of breaking down Raft/Paxos/etc. into the individual actions that comprise consensus. planetscale.com/blog...

To South Bay folks: South Bay Systems is currently limited by finding venues willing to host. If you know of somewhere that can host 60+ people and is happy getting a few minutes for a quick "here's who we are, what we do, and who to talk to for hiring" in return, please let me know!

If you'd like to do something great for the database community, write an easy to use implementation of the TPC-* benchmarks. Bonus points if it comes with blog posts describing what it is in a database that each TPC-* benchmark is actually testing.

@bernsteinbear.com Would you happen to have a suggestion for an easy instruction selection algorithm, sort of like the recommendation of DDCG for quick and easy register allocation?

A distributed systems reliability glossary jepsen x antithesis antithesis.com/resources/re...

The progress on standardizing vector support is indeed pretty cool to hear about peter.eisentraut.org/blog/2025/06...

The South Bay Systems meetup is back on July 22! We're excited to welcome Jacopo Tagliabue (co-founder and CTO of Bauplan) for a talk on Speedrunning the Lakehouse: Shipping a FaaS that Looks Like a Database (and Vice Versa). 🔗 lu.ma/5zcpbae9

Both instances of vendors reporting highly misleading TPC-C results have come from the use of sysbench-tpcc, so let's see if we can try to improve the root of the problem rather than one-by-one publicly shame vendors? github.com/Percona-Lab/... Also open to better ideas.

I ran across a pretty nice x86 SIMD intrinsics cheat sheet: db.in.tum.de/~finis/... Thanks Jan Finis!

Has anyone looked at using siso (pkg.go.dev/go.chromium....) outside of chrome, as an easy way to gain RBE-based distcc-like compilation?

github.com/panghy/javaf... is a rewrite of FoundationDB's Flow language/framework for concurrency and deterministic simulation rewritten into Java by one of FoundationDB's earliest users.

I'm getting tired of these new paper announcements on LinkedIn which are like

[arXiv] TreeTracker Join: Simple, Optimal, Fast arxiv.org/abs/2403.0... TreeTracker gives a very simple breakdown of what the core differences are between a naive binary join and an optimal multi-way join.

[EDBT'22] Integrating the Orca Optimizer into MySQL openproceedings.org/... Huawei Taurus folk plugged Orca into MySQL to get a better query optimizer for OLAP queries in MySQL.

Disaggregated OLTP pricing does not seem to be very consistent across cloud vendors. For 2 CPU, 8GB RAM, and no storage costs, I see: Alibaba PolarDB $40/mo Amazon Aurora $105.12/mo Huawei TaurusDB $143.40/mo Google AlloyDB $159.6672/mo Azure SQL DB $368.188/mo

Wisdom on btree prefix compression from someone who has implemented and benchmarked it. From lobste.rs/s/za4cxl/b...

Wrote up a little bit of meta-commentary about how I'd love to see more people blogging notes about academic papers they find useful simonwillison.net/2025/Jun/13/...