Ahsan Hadi: PostgreSQL 19 Is Here: Worth the Wait
PostgreSQL 19 is available through the pgEdge repository, with installation instructions for apt and dnf systems. Release notes and a blog series explain the project’s changes. (pgedge.com)
The claim that “big data” is the central obstacle to making better decisions has outlived its usefulness. For years, technology vendors warned that exploding data volumes would overwhelm traditional systems and urged organizations to adopt platforms built for massive scale. But data has not grown as dramatically as predicted, while hardware and conventional databases have improved. Drawing on more than a decade of experience building BigQuery and analyzing customer usage, Jordan Tigani argues that many organizations still struggle to use their data even after buying large-scale tools, because data volume was never the main problem.
The evidence suggests most organizations hold modest amounts of analytical data: the vast majority of BigQuery customers had less than a terabyte, and heavily active customers had a median well below 100 GB. Industry feedback put a typical warehouse around 100 GB; even large technology companies often had only a few terabytes, while many had far less. Data sizes follow a steep power-law distribution, with a few extreme cases masking the norm. Ordinary business activity also generates data slowly: a medium-sized company’s orders might take years to reach a gigabyte. The argument distinguishes storage growth from computing needs. Data accumulates over time, but most analysis concerns recent records, so compute requirements often stay relatively stable. Cloud separation of storage and compute makes retaining large archives practical without requiring equally large processing capacity. The practical priority, then, is using data effectively - not scaling for hypothetical volume.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
PostgreSQL 19 is available through the pgEdge repository, with installation instructions for apt and dnf systems. Release notes and a blog series explain the project’s changes. (pgedge.com)
Graham Dumpleton, creator of mod_wsgi, is building Muru, an AI-assisted Rust server for running Python’s WSGI and ASGI apps in development and production. Its features include HTTPS, Kubernetes support, scaling, and tracing, with Dumpleton using his two decades of experience to guide and refine the project. (simonwillison.net)
Fast-Anything helps optimize Python, Node, and Ruby apps by profiling them and rewriting only their slowest functions in Rust. It uses coding agents to make the changes and checks that the rewritten functions behave the same. (github.com)
Leonida is a community project that aims to recompile the Xbox version of Grand Theft Auto 6 for native PC execution. Its public GitHub repository is GPL-3.0 licensed and has attracted thousands of stars and hundreds of forks. (github.com)
Rust GUI development is gaining momentum as AI makes Rust easier to use and teams seek safer, faster alternatives to bloated cross-platform apps. New toolkit projects should focus on clear use cases and practical trade-offs instead of chasing a universal definition of “native.” (linebender.org)
Rex connects to exe.dev virtual machines by running a Rex server remotely and forwarding its Unix socket over SSH, keeping terminal sessions available across disconnects. Its shared session model lets users and agents manage remote terminal windows and splits without relying on tmux. (blog.exe.dev)
Today's best Hacker News stories, summarized and screenshotted, one email a day.