pgEdge

The main heap for a 100,000-row PostgreSQL table is 47 MB when TOAST moves the big JSON column out of line and 318 MB when it doesn't (heap only, so the toast table's own storage isn't in that number). TOAST, The Oversized-Attribute Storage Technique, is how Postgres keeps large values in a hidden companion table and leaves a pointer in the row. Shaun Thomas's latest PG Phriday starts with a slow event log he once fought with, and the twist is that the problem wasn't necessarily TOAST but the column itself. Counting rows ran in 26 ms against 91 ms and an UPDATE took 62 ms against 128 ms, because small rows pack densely. Pulling one field out of the JSON took 288,127 buffer accesses against 40,754, because every toasted value means a trip to the toast table. So what do you do when the big column is the problem? Splitting it into its own table behind a LEFT JOIN view dropped the main heap from 47 MB to 5 MB, and the planner skips the join for queries that never ask for the column. Lowering toast_tuple_target won't get you there, since rows under TOAST_TUPLE_THRESHOLD never get toasted. And lz4 compressed nearly 30 times faster than pglz in his test, on very repetitive data, so your ratio will vary. The queries and their output are in the full post: hubs.la/Q04CXnkH0 #postgresql #postgres #pgphriday #dba #database #performance #sql #compression #lz4 #devops #opensource #dataengineering #programming #tech

Should you go to KubeCon + CloudNativeCon North America? A quick checklist. Check any that apply: [ ] You run, or are about to run, stateful workloads on Kubernetes [ ] You’d like to be in a room with 9,000+ people building cloud native infrastructure, across 400+ sessions [ ] You want to meet the maintainers behind the projects you rely on (Maintainer Tracks and ContribFests run all week) [ ] You like a head start: Monday has co-located events like Platform Engineering Day, Observability Day and Cloud Native AI + Inference Day [ ] You’d rather not rent a car: the Salt Palace sits inside the downtown free fare zone, and TRAX gets you from the airport in about 20 minutes Checked even one? Put Salt Lake City, November 9-12, on your calendar. pgEdge will be there too, talking Postgres on Kubernetes with the cloud-native community. Come by the pgEdge table in the sponsor area and say hi. Register: hubs.la/Q04CV9kw0 #KubeCon #CloudNativeCon #PostgreSQL #DevCommunity #SaltLakeCity #Utah #TechConference #Tech #Kubernetes #CNCF #CloudNative #CloudEngineering #CloudOps #Containers #Docker

AI has made it easy to prototype an app in a weekend. Getting it into enterprise production is the hard part. In this clip, Phillip Merrick explains pgEdge Starfleet in a nutshell: a developer-friendly Postgres cloud platform that lets you build fast, then deploy where your data needs to live. That can be pgEdge-hosted cloud or your own cloud (AWS, Azure, or GCP). With pgEdge Enterprise Postgres, it can also be on-premises or air-gapped. It’s the same Postgres platform in each case, so you don’t have to replatform on the way to production. Two things set it apart: 👉 Deployment flexibility. Start in minutes on our cloud, then move to your cloud or your own infrastructure when compliance, sovereignty, or scale demands it. 👉 Postgres expertise. Starfleet is built by people with decades of Postgres experience, some going back to the early days of the open source project. It also ships with AI tooling built in: an MCP server for connecting tools like Claude Code, Cursor, and Replit, a RAG server, and a PostgREST API server. SafeSession keeps read-only MCP connections truly read-only. Starfleet is generally available now. You get a 14-day free trial with no credit card, and plans start at $25/month. Watch the clip, then try it: hubs.la/Q04CM9Z50 #dev #programming #appdev #applicationdevelopment #softwareengineering #softwareengineer #hacking #aiengineering #llm #llmops #devops #dba

AI has collapsed the time it takes to prototype an app, but the demands of production (compliance, scale, security) haven’t changed. That gap is where Gil Browdy, pgEdge’s Director of Engineering for cloud, started last week’s webinar on pgEdge Starfleet, our enterprise Postgres cloud platform... and you can still register and watch it on demand. The session follows the build. Starfleet runs standard community Postgres, with branching handled by copy-on-write at the file system layer instead of a replaced storage layer, so a branch spins up fast, gets its own connection string and MCP and RAG servers, and is billed only for the time it exists. Security is on by default: MCP servers are read-only out of the box, IP allow lists deny everything until you say otherwise, and pgEdge SafeSession keeps read-only connections read-only. Then Paul Rothrock drops a table on a branch live, with the production table left still sitting there. The Q&A covers restores, bringing over an existing Postgres database, and a good first branching workflow. Watch on-demand: 🎬 hubs.la/Q04CnhGC0 #postgresql #postgres #webinar #opensource #database #cloud #developers #databasebranching

A multi-master PostgreSQL cluster can report every subscription as healthy while the new node publishes nothing to the rest of the cluster. That’s one of the things Ahsan Hadi ran into while testing a rolling upgrade of a live PostgreSQL 17 cluster to 18 with Spock, pgEdge’s open-source extension for multi-master logical replication, following Susan Douglas' documentation on rolling upgrades. The approach: add a PostgreSQL 18 node, move traffic, retire the old node, repeat. In the test the application never lost a writable node. The catch is that none of the problems he hit showed up in a status check. His list of what to watch for: with DDL replication off, the new node’s tables sit in no replication set until you add them, the new node’s Spock slots need failover = true before slot sync will touch them, and Patroni overwrites Spock’s settings unless they’re in its etcd config. So test writes in both directions before you remove anything. A failover test is still on his list, which means a follow-up post is coming. Stay tuned... Read the full walkthrough from Ahsan and Susan: hubs.la/Q04Cct900 #postgresql #postgres #opensource #database #dba #highavailability #distributedsystems #patroni #devops #sre

PostgreSQL Conference Europe in Valencia is all wrapped up - and it was a fabulous event! The book signing turned the pgEdge table into a mob scene. :-) Our own Jimmy Angelakos was a rockstar signing copies of his Manning book, PostgreSQL Mistakes and How to Avoid Them, a tour of common PostgreSQL mistakes with a demo for each, in full costume through two boxes of books. The talks were packed all week, and our team gave several: Dr. Ibrar Ahmed on eight steps to database disaster recovery, Phillip Merrick on agentic engineering, operations and analytics with Postgres, and Jimmy on radar and radar-ui, agentless diagnostic snapshots for PostgreSQL, plus the On-Call Confessions panel. Andrei Lepikhov gave a talk too on “Finding the Invisible: Spotting Query Planner Issues Before They Appear”, introducing pg_track_optimizer, an extension he built to compare what the query planner expected with what actually happened. Thank you to PostgreSQL Europe for organizing, and to everyone who stopped at the pgEdge table in the sponsor area. It was a fun week. 🐘 Miss getting a copy of the book in person? You can still get it online: 📖 hubs.la/Q04C6ZXh0 #postgresql #postgres #queryplanner #dba #opensource #community #conference #databases #spain #programador #tecnologia #devcommunity #valencia

Someone runs UPDATE orders SET status = ‘cancelled’; and forgets the WHERE clause. Now what? For most Postgres teams, the answer is an afternoon of restore-and-reconcile: bring up a copy from before the mistake, pull out the damaged rows, and patch them into a live table that kept moving. It works. It also ties up your most senior person for hours. Muhammad Aqeel built pgVolvra to turn that into one statement. It’s an open-source tool for row-level undo and history in Postgres. It records changes on the tables you choose, then reverses only the rows a mistake touched. And if users changed some of those rows in the meantime? Undo stops and tells you which ones are in the way. There’s deliberately no force flag. It’s plain SQL, so no extension and no superuser, which is why it runs on managed Postgres. Aqeel verified it on Supabase, Neon, and pgEdge Starfleet, our Postgres cloud platform. There’s also an optional archive tier, plus replay for the hours a restore leaves behind. It’s a beta (Postgres 14 through 19), so reports of where it breaks are welcome. Check out the blog post: hubs.la/Q04C4_2F0 You can find the code, docs, and test suite on GitHub 🔗 hubs.la/Q04C4_1L0 #PostgreSQL #Postgres #OpenSource #DBA #DatabaseAdministration #DataRecovery #SQL #DevOps #DataEngineering #pgEdge

pgAdmin ships a new release roughly every six weeks, and until recently the machinery behind those releases was a Jenkins server that only a few people could get into. 🔎 pgAdmin is the open source graphical management tool for PostgreSQL, started by Dave Page nearly 30 years ago, and he still works on it as part of his job at pgEdge. His two new posts cover how its releases get built. The first blog describes the move of the entire build and release process to GitHub Actions, so it lives in the repository and goes through pull requests like any other code. The second covers pgbuild, which builds the libraries bundled into the desktop packages and pins a SHA-256 digest for each one, so a swapped archive fails the build. These engineering deep-dives explore a build config hidden in a web UI, “latest” tags that shift underneath you, and unverified dependencies can bite any project that ships packages, images or binaries. Dave walks through the fix, in the open. Part one, the build farm: 📖 hubs.la/Q04BQHlL0 Part two, pgbuild: 📖 hubs.la/Q04BQHnl0 The project: 🔗 hubs.la/Q04BQHrP0 and hubs.la/Q04BQHgn0 #postgresql #postgres #pgadmin #opensource #githubactions #cicd #devops #programming #engineering #deepdive #tech #developer #devcommunity #database #dba #softwareengineer

Phillip Merrick, pgEdge’s CEO and Co-Founder, speaks at PostgreSQL Conference Europe in Valencia this morning, 11:45 to 12:35 local time in SM1-A: “Agentic Engineering, Agentic Operations and Agentic Analytics with Postgres.” The talk looks at how AI agents can help across the whole Postgres lifecycle: building applications, diagnosing database problems, and analyzing years of historical data. We've addressed all three situations through fully-open source tools (under the PostgreSQL license): The Agentic AI Toolkit covers engineering, the AI DBA Workbench covers operations and administration, and pgEdge ColdFront (currently in public beta) covers analytics. (🔗 Want to check these out ahead of time? Visit our GitHub: hubs.la/Q04BPYRW0) We’re also giving away free signed copies of PostgreSQL Mistakes and How to Avoid Them, Jimmy Angelakos’s 2024 Manning book on the mistakes Postgres users tend to make and how to sidestep them. Come on by the booth anytime to meet the author and get a copy 📖 Session details: hubs.la/Q04BPYQ30 #postgresql #postgres #agenticai #aiagents #valencia #spain #programador #tecnologia #devcommunity #techconference #tech #programming

An application built quickly with AI coding tools still has to run somewhere that meets the enterprise's security, compliance, and reliability requirements. 🔒 That's the gap Phillip Merrick, co-founder and CEO of pgEdge, takes on in CTO Magazine. On where these apps end up running, his point is that "enterprise data often can’t leave the enterprise." Gizel Gomes' piece turns that into a checklist for CTOs: weigh security, compliance, and data needs before choosing where an AI-built app runs (a vendor-managed cloud, your own cloud account, or on-premises), measure how often an AI agent gets things wrong and decide whether that rate is acceptable, steer teams toward vetted AI coding and database platforms before shadow IT does the choosing, and give every AI-built application a long-term owner. pgEdge Starfleet, our enterprise Postgres cloud platform, is built around the same view: start on pgEdge-hosted cloud, then deploy to your own cloud or on-premises with pgEdge Enterprise Postgres, without replatforming. Read the full piece in CTO Magazine: 📖 hubs.la/Q04BHyxH0 #postgresql #postgres #cto #enterprise #ai #aicoding #security #compliance #datasovereignty #shadowit #governance #opensource #database #devops

We're in Valencia for PostgreSQL Conference Europe 2026, and day one is underway. New to Valencia (or the conference)? The conference website covers more than the schedule - it has venue and travel info, how lightning talks work, details on the social event, ideas for things to do around Valencia, and more info about the Community Events Day on Friday. Check it out if you haven't already 👀 Four pgEdge sessions are on the program this week. Jimmy Angelakos is on the Postgres On-Call Confessions panel (Tue 13:45, Audit 1), Dr. Ibrar Ahmed walks through eight steps to database disaster recovery (Tue 14:45, SM1-B), Phillip Merrick talks agentic engineering, operations and analytics with Postgres (Wed 11:45, SM1-A), and Jimmy is back Thursday with radar and radar-ui, agentless diagnostic snapshots for PostgreSQL (10:25, Audit 1). All times are Valencia local. We're a proud Gold Sponsor of the event. Don't forget to come by the pgEdge table in the sponsor area. If hosted Postgres for enterprise requirements is interesting to you, we'd like to hear what you're dealing with. pgEdge Starfleet hosts 100% standard PostgreSQL on pgEdge-hosted cloud or in your own AWS, Azure or GCP account, with your choice of deployment at any point. 👉 Learn more about the conference: hubs.la/Q04BHNQG0 #postgresql #postgres #valencia #spain #espana #basededatos #tecnología #programador #opensource #database #techevent

Spinning up a PostgreSQL database for a new app takes minutes. Finding out later that the platform can't follow the app into production is the slow part. 💫 That's the problem behind today's webinar, live at 8 AM PT (11 AM ET): "Build on Cloud, Deploy Anywhere: pgEdge Starfleet Deep Dive + Live Demo." pgEdge Starfleet is an enterprise Postgres cloud platform built on 100% open source community Postgres, and the hosted side is built for the fast start. Paul Rothrock (Solutions Engineer) demos sign-up to a running database in under two minutes, and the session covers true copy-on-write branching, with fully isolated database branches created in just a few minutes. Gil Browdy (Director of Engineering, Cloud) talks about how it fits together once you're ready to move: from pgEdge-hosted cloud to your own AWS, Azure or GCP account, or on-premises and air-gapped with pgEdge Enterprise Postgres, without replatforming. The live Q&A at the end - don't forget to bring your questions to the session. This is happening today at 8 AM PST // 11 AM EST. Register now to join live or watch on-demand later: 🎙️ hubs.la/Q04BHJnH0 #techwebinar #techpodcast #webinar #postgresql #postgres #developers #appdev #cloud #database #opensource #devtools #devcommunity

Where your PostgreSQL database runs is an easy decision on day one and a harder one when the application has to move. 🐘 Tomorrow, Oct 20 at 8 AM PT (11 AM ET), pgEdge hosts "Build on Cloud, Deploy Anywhere: pgEdge Starfleet Deep Dive + Live Demo." Gil Browdy (Director of Engineering, Cloud) covers how pgEdge Starfleet, an enterprise Postgres cloud platform built on 100% standard, open source community Postgres, works. Paul Rothrock (Solutions Engineer) then demos the path from sign-up to a running database in under two minutes. For developers, that means starting on pgEdge Starfleet Managed, then deploying to your own AWS, Azure or GCP account, or on-premises or air-gapped with pgEdge Enterprise Postgres, without replatforming. True copy-on-write database branching is on the agenda too. The live Q&A at the end is with Gil Browdy, one of Starfleet's key architects, so come with questions. Register now to join live or watch on-demand later: 🎙️ hubs.la/Q04BChxB0 #postgresql #postgres #webinar #developers #appdev #backend #cloud #database #opensource #devtools #devcommunity #techcommunity #techwebinar #techpodcast #podcast #tech

"It appears that agents are now spinning up more Postgres databases than humans." That's one line from pgEdge's sponsor interview with PostgreSQL Conference Europe, now live on the conference site. We talk about where we see Postgres growing (agentic platforms, and enterprises moving off legacy databases), why an open source license matters when you need to inspect the technology you depend on, and what regulated industries and government ask of a database: compliance, security, high availability, and flexible deployment. The interview also covers the people behind it: a remote-first, globally distributed team that includes a Postgres core team member, significant contributors, and board members of PgUS, PGCA, and PGEU. Read the interview: 🎙️ hubs.la/Q04Bt3pY0 PGConf EU 2026 runs October 20-23 in Valencia, Spain, and pgEdge is a Gold Sponsor. Haven't registered yet? You can do that here: hubs.la/Q04Bt3rz0 Already on your way? Come by the pgEdge table in the sponsor area and say hello. We'd love to hear what you're building. #PostgreSQL #Postgres #OpenSource #PGConfEU #Valencia #Databases #AgenticAI #DevOps #DBA #OpenSourceDatabases

“The interesting part of a build system is not how clever it is but how much of it you can see.” That’s Dave Page, creator of pgAdmin, closing out a two-part series on how pgAdmin is built. This second post covers pgbuild, the repository that compiles PostgreSQL, OpenSSL, Kerberos and other libraries bundled into pgAdmin’s Windows and macOS packages. Its releases sat under tags that were replaced nightly, and there was no checksum anywhere in the chain. Dave mentions that the build process was “not a gaping hole so much as an unnecessary second door into the same room.” A signature vouches for what’s inside a package, and with no checksums the project couldn’t prove the libraries were the ones it intended. The fix runs in both repositories. pgAdmin now records the SHA-256 of every archive it consumes and fails the build on a mismatch, and pgbuild publishes an immutable release for each upstream version to pin against. So which OpenSSL and which libpq are in the installer you downloaded? The project has been asked that before and, in Dave’s words, answered “with more hedging than I was comfortable with.” It can now say precisely which version went into which release. Read the full piece: 📖 hubs.la/Q04BrDNK0 #postgresql #postgres #pgadmin #opensource #softwaresupplychain #supplychainsecurity #devops #githubactions #pgedge

“For just about any data management problem, the answer is Postgres.” That’s Phillip Merrick, pgEdge’s co-founder and CEO, in Paul Gillin’s new SiliconANGLE piece on why a 40-year-old database is getting fresh attention as AI coding agents write more of the applications. Phillip’s point is that agent-driven development is multiplying the number of databases to manage. He calls Postgres “the default database for agent-generated applications.” In his words: “You’ve only got a fixed number of human [database administrators].” The article also covers what keeps that manageable: read-only access enforced at the engine level, access controls on how agents reach the data, and a human approving any fix an AI suggests. If you’re deciding where agents get write access in your own stack, the reporting on those guardrails is a useful starting point. Read the full piece: 📖 hubs.la/Q04Bpypj0 #postgres #postgresql #ai #aiagents #database #dba #opensource #developers #devops #llm #pgedge

Dave Page, creator of pgAdmin - the PostgreSQL admin tool he started nearly 30 years ago - has moved the build farm for the project from Jenkins to GitHub Actions. The old setup was jobs configured through a web interface, with Dave as the person who knew how it was wired together and fixed it when it broke. Now the Debian and Red Hat packages, Windows installer, macOS app bundle, Python wheel, container image, Helm chart and docs are all built from workflows in the repo, so a packaging change shows up in a pull request like any other. Before anyone can promote a release candidate, an acceptance test installs it the way a user would: from the APT and YUM repos on nine distributions with no --nogpgcheck anywhere, plus signature checks on the Windows and macOS installers. Building the test surfaced two bugs: packages published one directory deeper than the repo metadata claimed, and noarch RPMs filed where no dnf client would look. Every file existed and every signature validated... but the repos didn’t work. Windows signing and publishing still can’t run on hosted runners (a hardware token and SSH keys, respectively). Both use self-hosted runners and are now documented in the repo. Read Dave’s full write-up: 📖 hubs.la/Q04BlJ9y0 #pgadmin #postgres #postgresql #opensource #githubactions #cicd #devops #buildengineering #packaging #softwaredevelopment #developers

pgBackRest, largely considered one of the most widely used PostgreSQL backup tools, is getting its own session at PostgreSQL Conference Europe 2026 in Valencia, Spain. 🐘 "The Monday We Learned Who Maintains Our Backups" is on Tuesday 20 October, 14:45 local time, in Audit 2 (Beginner level). David Steele (pgBackRest) and Jan Wieremjewicz (Percona) cover why the tool was created, how it became a core part of the PostgreSQL ecosystem, what maintaining it involves, and how users, vendors and the community responded when doubts about its future spread. The close is in regards to funding and succession planning. New to it? pgBackRest is a free, MIT-licensed backup and restore solution for PostgreSQL. It supports parallel backup and restore, full, differential and incremental backups at the file or block level, per-file checksums rechecked on restore, repository encryption, and S3, Azure or GCS compatible object storage. pgEdge is one of pgBackRest's sponsors, and a Gold Sponsor of the conference. If you're in Valencia, come by the pgEdge table in the sponsor area. Session details: 🎙️ hubs.la/Q04Bj0970 #postgresql #postgres #pgbackrest #backup #disasterrecovery #opensource #database #dba #valencia #spain #techevent #techcommunity

PostgreSQL Conference Europe 2026 starts next Tuesday in Valencia, Spain, and registration is still open. 🐘 If production failure stories are your thing, Tuesday, October 20 is the day to be in the room (all times are Valencia local time): 🎙️ 13:45, "Postgres On-Call Confessions: Worst Practices 101," a six-person panel with pgEdge's Jimmy Angelakos. It ends with an open mic, so bring your own production "obvious fix" that backfired to contribute to what should be quite the entertaining discussion. 🛟 14:45, Dr. Ibrar Ahmed on "Eight Steps to Bullet-Proof Database Disaster Recovery." Every command in the talk runs on standard PostgreSQL, no third-party tools required, to help you have a procedure in place when systems go down. Wednesday and Thursday belong to Phillip Merrick on agentic engineering, operations, and analytics with Postgres (11:45 Wednesday), as well as Jimmy who is giving a talk on radar, an agentless diagnostic snapshot tool for PostgreSQL (10:25 Thursday). pgEdge is a Gold Sponsor, so come by the pgEdge table in the sponsor area and say hi, whether you're coming from Spain or from across Europe. 👋 Thinking about attending? Don't forget to register: 🔗 hubs.la/Q04BbgYT0 #postgresql #postgres #valencia #spain #europe #techevent #techcommunity #devcommunity #database #dba #opensource #disasterrecovery