Is this you? As a journalist, you can create a free Muck Rack account to customize your profile, list your contact preferences, and upload a portfolio of your best work.
Claim your profile
Articles by Matty Stratton
You Didn't Choose Your Split Architecture. You Accumulated It.
Ask a team when they decided to run a second database and nobody remembers the meeting, because there wasn't one. A slow report, a cautious feature launch or an org boundary adds a system, and each step looks reasonable on its own. The cost is structural: a second copy isn't the source of truth and isn't current when you query it, however good the pipeline is. Sometimes a separate system is the right call. It should still be a decision.
Most teams evaluating a time-series database are really asking how much of their data to move out of Postgres. For most of them, the time-series side is most of the data. That question comes from an era when that data had to live in a separate system. Postgres with TimescaleDB doesn't require the split. The payoff is that every decision you make after this one gets smaller.
It's Been Ten Years of ADO, Charlie Brown
It's been ten years of Arrested DevOps! Joe, Matty, Bridget, Jess, and Trevor spend some time (quite a lot of time!) reminiscing over stories and history of the podcast.
Why We Aren't Ready for C-3PO
When people picture a robot, they picture C-3PO (or if you're a Trekkie, it's Lieutenant Commander Data and his brother Lore): a humanoid that walks into any room, understands what you say, picks up anything, and improvises when things go sideways. Decades of science fiction trained us to treat that as the target, the thing we are all building toward and almost have. For years the working assumption was that we were waiting on smarter robot brains. The evidence points the other way.
Don't Give the Model SQL
I built a web app to answer questions about my own health data, for a reason that is embarrassingly small. I already had a perfectly good way to ask them: a Claude Code skill that queried the database and reasoned over the results. It worked well. It also can't run on Claude iOS, and the moment I actually want to ask "should I be worried about this" is standing in a kitchen at 6am, not sitting at a desk.
Every Number I Didn't Measure Was Wrong
I went to publish something about the database behind my personal health tracker, and fact-checking it took a day. By the end, most of the numbers I had written down to justify that database's design had turned out to be wrong. Not wildly wrong. Wrong in the quiet way, where the conclusion still holds and the reasoning under it has rotted.
The Past Keeps Changing
I moved my personal health tracker onto a time-series database this month. It holds 69,000 rows in thirty megabytes. That is not a defensible reason to use one, and I want to be blunt about it before you get there on your own. Nothing I did made anything faster and nothing needed to. What it replaced was smaller still. An 840 KB SQLite file on my laptop, seventeen metrics, about 280 health observations going back three months, plus two and a half years of lifting history.
The Data Layer for the AI Data Center
This is Part II of a two-part series on the AI data center stack. Part I, AI's Physical Constraints: How AI Rewired the Data Center, explains why AI capacity has become constrained by the physical data center stack: accelerators, memory, cooling, power, time, and water. Physical Plant Telemetry Is Now a Data-Layer Requirement In an AI data center, the workload no longer stops at the server boundary.
I Contain Multitudes (and Also Three Git Repos)
My blog is older than most of the tools I use to run it. The oldest posts started life on LiveJournal sometime in the early 2000s, got dragged through a WordPress import at some point in the mid-2000s, spent the better part of a decade as a Hugo site, and, about a week ago, moved into an Astro 5 site that reproduces every one of those old URLs byte for byte. That's 2,630 posts spanning 2001 to 2020, frozen exactly where they landed.
Your Agent Didn't Break Prod. Your Pipeline Did.
Picture a pipeline that looks pretty reasonable on paper. An agent opens pull requests. CI gates the merge to main: lint, tests, build, all green or it doesn't land. A scheduled job periodically promotes whatever's sitting in staging straight to production. No individual approval per feature. Just a batch cutover on a timer. That pipeline will absolutely, eventually, wreck your week. Not because the agent did anything malicious.
AI's Physical Constraints: How AI Rewired the Data Center
For most of the cloud era, a server rack was a five to twenty kilowatt object. You could fill a room with them, move air across the front, and the building stayed an ordinary building. A single current AI rack, NVIDIA's GB300 NVL72, draws about 132 to 140 kilowatts, with the GPUs alone accounting for more than a hundred. That is close to an order of magnitude more power in the same floor space as those old racks, and it lands as heat in the same small volume.
What We Heard: Three Patterns from Spring 2026 Events
Spring 2026 was relentless. Tiger Data teams covered ten events across Europe and the US: GrafanaCON in Barcelona, Hannover Messe and AWS Summit Hamburg in Germany, AWS Summit London, ETHDenver in Colorado, the Offshore Technology Conference in Houston, Data Driven Oil & Gas in Texas, Sensors Converge and IOT Tech Expo in the Bay Area, and AWS Summit Los Angeles.
What's Actually in My .zshrc (and Why)
Your shell config is mostly a graveyard. A line you added in 2019 to fix a problem you don't remember having. An alias you're afraid to remove because what if you need it. Mine is no different, but there are a handful of pieces in there that are worth explaining, not because they're clever, but because most people have never thought about why they'd want them. This isn't a full dump of my config. If you want that, it's all on GitHub at matty-dotfiles.
What You’re Really Owning When You Self-Host TimescaleDB
Why operating TimescaleDB for mission-critical applications becomes a sustained platform engineering investment. Written by the team that builds TimescaleDB and operates Tiger Cloud across thousands of production deployments. Most engineers who evaluate TimescaleDB believe they are making a database decision. By the time customers depend on the application in production, they discover they made a platform ownership decision. The gap between those two decisions is the subject of this paper.
The Engineering Calendar Is the Database Bill Nobody Tracks
The database bill went up 40% last quarter, and everybody noticed. Finance noticed. You had a meeting about it. Somebody made a slide with a trend line that looked bad enough to earn its own agenda item. The other database bill probably did not make the deck.
The True Cost of Database Optimization: Engineering Time
Copy as HTML Open in ChatGPT Open in Claude Open in v0 DatabasePostgreSQLPostgreSQL Performance Copy as HTML Open in ChatGPT Open in Claude Open in v0 "We can fix the performance issue with better indexes, smarter partitioning, and some vacuum tuning. It's cheaper than switching." You've heard this sentence. You may have said this sentence. The optimization wasn't cheap. It just felt like it was. "Cheaper than what" is the question nobody asks. The optimization doesn't show up on an invoice.
Five Warning Signs Your Database Needs Different Architecture
Copy as HTML Open in ChatGPT Open in Claude Open in v0 Copy as HTML Open in ChatGPT Open in Claude Open in v0 Every database has bad days. A slow query after a schema change. A spike in replication lag during a traffic surge. An autovacuum job that runs long enough to make you nervous. Those are tuning problems. They have tuning solutions. A senior engineer can fix them in a day, sometimes in an hour, and the fix holds. You move on. But there's a different category of symptom. The kind that comes back.
Autovacuum: The Tax You're Always Paying
Copy as HTML Open in ChatGPT Open in Claude Open in v0 Copy as HTML Open in ChatGPT Open in Claude Open in v0 You open pg_stat_activity during a write peak and there it is. Autovacuum. Again. You've tuned this three times. You wrote a runbook for it. At this point it has its own section in the quarterly database review. The table it's working on hasn't seen a single UPDATE in six months. Every write is an INSERT. Old data gets dropped by partition, not deleted row by row.
Postgres Extensions Cheat Sheet: Replace 7 Databases With SQL
Copy as HTML Open in ChatGPT Open in Claude Open in v0 Table of contents 01 Enable Everything06 Caching (Replace Redis)08 Documents (Replace MongoDB)12 What's Next Table of contents 01 Enable Everything06 Caching (Replace Redis)08 Documents (Replace MongoDB)12 What's Next Copy as HTML Open in ChatGPT Open in Claude Open in v0 This post is a practical companion to It's 2026, Just Use Postgres. That post makes the architectural case for consolidating on Postgres. This one shows you how.
Why I'm Learning ROS 2 as a Database Person
There's a moment that happens in every robotics company that makes it to production. The pilot worked. The robot does the thing. Now there are twenty on the floor, then fifty, then a fleet, and somebody asks: "What did the sensors look like on unit 17 during those three anomalies last quarter?" That's when the data infrastructure question gets real. As robotics moves from lab to warehouse floor to factory line, that question is going to come up a lot more often.
Optimization vs. Architecture: Knowing the Difference
Copy as HTML Open in ChatGPT Open in Claude Open in v0 Table of contents 01 What optimization can fix03 The recurrence test06 The decision Table of contents 01 What optimization can fix03 The recurrence test06 The decision Copy as HTML Open in ChatGPT Open in Claude Open in v0 There are two kinds of database performance problems. The first kind responds to optimization. You add an index, queries speed up, the improvement holds. Done. The second kind responds to optimization temporarily.
How My Coworker Who Didn't Know 'cd' Shipped to Production
This morning, our Design Lead asked me how to get her terminal into the right folder. Tanya [6:57 AM] my ghost isnt in tiger den anymore! GAH do i do /tiger-den or what was the command to make it work in tiger den Matty [7:14 AM] What what? Tanya [7:14 AM] sorry lol you know when i go into terminal/or ghostie and i need to type the thing to make it so im working in tiger den Matty [7:14 AM] Yes type cd tiger-den Tanya [7:15 AM] YES OK CD thank you Tanya has been at Tiger Data for about two weeks.
Write Amplification in Postgres: The 3-4x Tax on Every Insert
Copy as HTML Open in ChatGPT Open in Claude Open in v0 Table of contents 03 Finding the floor04 The asymmetry05 The point Table of contents 03 Finding the floor04 The asymmetry05 The point Copy as HTML Open in ChatGPT Open in Claude Open in v0 You're inserting 100 MB/sec of sensor data into Postgres. Your monitoring dashboard, meanwhile, is cheerfully reporting 300-400 MB/sec of disk writes. You check the math twice. You check it a third time. The math does not improve.
ClickHouse Is Fast. Your Pipeline Isn't.
Copy as HTML Open in ChatGPT Open in Claude Open in v0 Copy as HTML Open in ChatGPT Open in Claude Open in v0 ClickHouse is fast. The benchmarks aren't lying. If you've run a comparison against vanilla Postgres on the same dataset, the results aren't close. ClickHouse wins by 10x-100x on typical analytical patterns. That benchmark is also only measuring one dimension of your decision. It tells you how fast queries run on static data.
The Best Time to Migrate Was at 10M Rows. The Second Best Time Is Now.
Copy as HTML Open in ChatGPT Open in Claude Open in v0 Copy as HTML Open in ChatGPT Open in Claude Open in v0 There's a pattern that plays out across almost every team running high-volume append workloads on vanilla Postgres. I've watched it happen enough times that I can practically set a timer. At 10M rows, everything is fine. Queries are fast. The team is shipping features. Nobody is thinking about the database. At 50M, queries start getting slow. Someone opens a Jira ticket about dashboard latency.
Read Replicas Don't Solve Write Bottlenecks
Copy as HTML Open in ChatGPT Open in Claude Open in v0 Table of contents 03 The replica lag problem05 Isolation vs. solution07 Where this leaves you Table of contents 03 The replica lag problem05 Isolation vs. solution07 Where this leaves you Copy as HTML Open in ChatGPT Open in Claude Open in v0 You added a read replica and something real happened. Dashboard queries stopped competing with ingestion. The primary's CPU dropped. Write latency came down. You watched the metrics and thought: that worked.
Document Databases: Be Honest
Copy as HTML Open in ChatGPT Open in Claude Open in v0 Copy as HTML Open in ChatGPT Open in Claude Open in v0 MongoDB gets a bad reputation in certain engineering circles that it doesn't entirely deserve. It ships fast. Schema flexibility is real. The developer experience for document-shaped data is good. A lot of teams made a reasonable call when they chose it. But there's a version of this story that ends badly, and it follows a recognizable pattern. The team picks MongoDB for a new system.
Postgres Performance: Why Peak Throughput Benchmarks Miss the Real Problem
Copy as HTML Open in ChatGPT Open in Claude Open in v0 Copy as HTML Open in ChatGPT Open in Claude Open in v0 You ran the benchmark. 80,000 inserts per second. The database handled it clean, latency stayed flat, no alarms. You shipped with confidence. Three months later, p95 write latency is creeping. Six months later, autovacuum is in your top processes by CPU. Nine months later, you're rebuilding indexes on a table that's crossed 400 million rows. The benchmark wasn't wrong.
MVCC: The Feature You're Paying For But Not Using
Copy as HTML Open in ChatGPT Open in Claude Open in v0 Copy as HTML Open in ChatGPT Open in Claude Open in v0 Most engineers have a working mental model of MVCC. Readers don't block writers. Concurrent transactions see consistent snapshots. It's why Postgres handles mixed read/write workloads so well, and it's a genuine engineering achievement. What's less obvious is that MVCC isn't free. Every row in every table carries its overhead. Not just rows that get updated.
Copy as HTML Open in ChatGPT Open in Claude Open in v0 Copy as HTML Open in ChatGPT Open in Claude Open in v0 Your system writes data constantly. Not in jobs. Not in batches. A stream that runs at 3am the same as it runs at 3pm. IoT sensors. Trade feeds. Metrics collectors. The data never stops. For a while, Postgres handles it fine. Then you start noticing things. Autovacuum is always running. Write latency has a pattern you can't explain by traffic alone.
Why Adding More Indexes Eventually Makes Things Worse
Copy as HTML Open in ChatGPT Open in Claude Open in v0 Table of contents 04 The feedback loop05 The diagnostic questions07 The bottom line Table of contents 04 The feedback loop05 The diagnostic questions07 The bottom line Copy as HTML Open in ChatGPT Open in Claude Open in v0 The pattern is familiar. A query is slow. You run EXPLAIN and see a sequential scan. You add an index. The query drops from seconds to milliseconds. You do this a dozen times over two years and it works every time.
The Hidden Costs of Table Partitioning at Scale
Copy as HTML Open in ChatGPT Open in Claude Open in v0 You implemented partitioning six months ago. DROP TABLE replaced your multi-hour DELETEs. Retention is clean. Queries got faster. It worked. Then something changed. Quarterly reports started slowing down. A new engineer broke the partition gap monitor. The pg_partman job failed silently over a weekend. You added a column and had to touch 400 tables. Partitioning solved the problem it was designed to solve. These are tradeoffs, not bugs.
Coding Agents Are Actually Good at This One Thing
The discourse around AI coding tools tends to collapse into two camps: people who think they're going to replace developers, and people dunking on "vibe coding" demos that fall apart the moment you look at them sideways. Both camps are mostly arguing about the wrong thing. I've been using coding agents heavily for the past few months, and the use case where they've actually changed how I work isn't production applications or greenfield SaaS ideas. It's internal tooling.
Vertical Scaling: Buying Time You Can't Afford
Copy as HTML Open in ChatGPT Open in Claude Open in v0 Copy as HTML Open in ChatGPT Open in Claude Open in v0 Your Postgres database is struggling. Write latency is climbing, autovacuum is fighting for I/O, and the indexes you added three months ago aren't cutting it anymore. So you do the obvious thing. You upgrade the instance. Metrics drop. Everyone exhales. Six months later, you do it again. Nobody puts this in a postmortem, because vertical scaling works. That's why teams keep reaching for it.
Understanding Postgres Performance Limits for Analytics on Live Data
Copy as HTML Open in ChatGPT Open in Claude Open in v0 PostgreSQLAnalyticsPostgreSQL Performance Copy as HTML Open in ChatGPT Open in Claude Open in v0 The Pattern Recognition Moment You're reviewing monitoring on a normal workday. There hasn't been a new deployment, no weird traffic spike, and no schema changes. But p95 write latency has crept from 8ms to 25ms over the past month, and last week it touched 45ms. Your largest tables crossed 500M rows sometime in March and they're still climbing.
Six Signs That Postgres Tuning Won't Fix Your Performance Problems
Copy as HTML Open in ChatGPT Open in Claude Open in v0 Table of contents 03 Data Is Append-Only06 Growth Is Sustained07 What to Do With This Table of contents 03 Data Is Append-Only06 Growth Is Sustained07 What to Do With This Copy as HTML Open in ChatGPT Open in Claude Open in v0 You've added indexes. You've partitioned tables. You've tuned autovacuum within an inch of its life. Performance improves for a few months, and then the dashboards go red again. Sound familiar?
Product 101: Your Secret Weapon for Understanding the Business
This is Part 5 of my 7-part series on business literacy for DevRel. Start with Part 1 if you missed it. We've covered sales, marketing, and finance. But there's one group we haven't talked about yet - and they might be your most valuable partnership in the entire company: Product. Not because product managers are inherently better than anyone else (though the good ones are pretty great). But because strategic product leaders sit at the intersection of everything.
Product-Led Growth and What It Means for DevRel
This is Part 5 of my 6-part series on business literacy for DevRel. Start with Part 1 if you missed it. We've covered sales, marketing, and finance - the three core business functions you need to understand. But there's one more concept that ties everything together, and it's especially critical if you work at a developer tools company or any company with a self-serve product: Product-Led Growth (PLG).
Finance 101: Budgets, P&Ls, and the Language of Money
This is Part 4 of my 6-part series on business literacy for DevRel. Start with Part 1 if you missed it. Time for everyone's favorite topic: money. Specifically, how companies think about money, track money, and make decisions about money. If your eyes are already glazing over, I get it.
Putting It All Together: Speaking Business as a DevRel Professional
This is Part 6 of my 6-part series on business literacy for DevRel. Start with Part 1 if you missed it. We've covered a lot of ground in this series. Sales pipelines and forecast accuracy. Marketing funnels and MQLs. P&Ls and CapEx vs OpEx. Product-Led Growth and how it changes the game for developer tools companies. If your brain feels a little full right now, that's normal. This is a lot of new vocabulary and new ways of thinking. But you now have the foundation.
Marketing 101: Funnels, Campaigns, and What Marketing Actually Means
This is Part 3 of my 6-part series on business literacy for DevRel. Start with Part 1 if you missed it. Let's talk about the relationship between DevRel and marketing. And if you just shuddered uncontrollably, you're exactly who I'm writing this for. A lot of people believe really hard that DevRel is not developer marketing, because somehow a lot of folks decided that sales and marketing are icky.
Sales 101: What Your Sales Team Does (And How DevRel Fits In)
This is Part 2 of my 6-part series on business literacy for DevRel. Start with Part 1 if you missed it. Let's talk about sales folks. And I need you to stick with me here, because I know what you're thinking. You've got a mental image, don't you? The stereotypical sales rep ... smooth-talking, coin-operated, willing to promise anything to close a deal. Someone who doesn't understand the product and just wants to hit their quota so they can get their commission check. Yeah, that person exists.
Why Business Literacy Matters for DevRel (And Why You Can't Skip This Step)
I hear over and over again, "the business doesn't understand DevRel." I'm here to tell you that is backwards - the problem is that DevRel doesn't understand business. Look, I get it. You didn't get into DevRel to spend your days talking about pipeline contribution and EBITDA. You got into this because you genuinely care about developers, you love building communities, and somewhere deep in your soul, you believe that if you make developers' lives better, good things will happen for everyone.
Fast, Zero-Copy Database Forks: Deploy on Fridays with Confidence
Your coding agent just shipped what looks to be a perfect migration. All tests pass. The schema refactor is elegant. It's Friday at 4pm. Are you brave enough to deploy? Most developers aren't. And for good reason. That AI-generated migration hasn't touched real production data. You don't know about the edge cases lurking in 40 million rows. You can't predict if the backfill takes 5 minutes or 5 hours. And you definitely don't know about that legacy integration from 2019 storing user IDs in a weird format.
How to Train Your Agent to Be a Postgres Expert
Agents are the new developer. But theyâre generalists. What happens when they design your Postgres database? Your schema runs, your tests pass⌠and six months later your queries crawl and your costs skyrocket. AI-generated SQL and database schemas are almost right. And thatâs the problem. Fixing schema design mistakes is costlier than refactoring code. It often means multi-week migrations, downtime windows, rollback plans, and your CEO asking why the site is in maintenance mode.
From Bureaucracy to Brilliance: How DevOps Culture Can Transform Cloud Governance
The Governance Theater Problem Imagine this - your CTO announces a major cloud governance initiative. Super critical to the business, all hands on deck, yada yada yada. Six months later, you have a 60-page policy document, monitoring dashboards generating 1,000+ alerts per week, and a SharePoint site full of standards that developers have bookmarked but never actually read. Sound familiar? This is governance theater - a classic example of "work as imagined vs. work as done".
Check if you are breaking your admin rules in your GitHub repos
Branch protection rules in GitHub are powerful! We can set rules that don't allow folks to commit directly to main, as well as enforce that PRs need to be reviewed and approved, and have required checks before merging. However, often we allow admin folks on the repo to be able to bypass these restrictions (the infamous "Merge without waiting for requirements to be met (bypass rules)" checkbox), which can be useful - checks can get stuck, sometimes we do need to do an emergency fix, etc.
A Brief DevOps History: Databases to Infinity and Beyond, Part 2
This is the second of a two-part series. Read Part 1. We left off in 1979, with the release of INGRES and Oracle v2. At that point in history, databases were almost exclusively a tool built for and used by enterprises, developed according to their needs. But then the 1980s arrived (in our opinions, the most fashionable decade) and with them came the desktop computing era and some historically significant hacker movies to go with them.
A Brief DevOps History: Databases to Infinity and Beyond
This is the first in a two-part series. The further we get from the origin of a piece of technology, the easier it is for us to take for granted the incredible amount of work and huge leaps in innovation that were required to get us to where we are today. Databases are one such piece of technology. Many of us were born decades after the first database was implemented, and while we know the technology is old, we have no understanding of the path taken to get where we are today.
Kat Cosgrove
Will Elon Musk actually destroy Twitter? CSAM poster re-instatement, “absolute free speech,” talent-gutting, massive advertiser exodus … What does it take to kill the bird? Yup. He borked it into a death spiral the instant he signed on as CEO. Essentially, yes. It’ll take a while to descend into its radioactive Xesspool destiny, but it’s inevitable it will join the ranks of 4chan, 8chan, Nchan, Xchan … Please yes, and quickly. I’m just done with all this stupid drama over this stupid little app.
Show More
loading
Actions
Is this you?
As a journalist, you can create a free Muck Rack account to customize your profile, list your contact preferences, and upload a portfolio of your best work.Get in touch with Matty
Contact Matty, search articles and posts on X, monitor coverage, and track replies from one place.
Learn more about Muck Rack