AWS Acquires DuckLabs
ducklabs.com869 points by onderkalaci 8 hours ago
869 points by onderkalaci 8 hours ago
I'm glad that DuckDB has a foundation in place and hope it is resilient enough to push the DB forward when the time comes.
Out of all the big orgs, Amazon is probably the one that has the least regard for keeping technically interesting projects alive, and the certainly will bulldoze it for some dumb reason when the next re-org comes.
I think 'least regard' belongs to Oracle. For some reason I thought DuckDB was one of Gabe's ideas over at DuckDuckGo :-).
That said, the acquisition by Amazon is interesting in that it seems to overlap with their existing DB offerings? Which makes me wonder what exactly it is that they are getting out of this.
> Amazon is probably the one that has the least regard for keeping technically interesting projects alive
Well ...
Either way that news is awful. Europeans selling out to US corporations and then wondering why Europe is not competitive in software engineering anymore. Funniest thing is Draghi keeps on saying that; well, recently we also heard that the USA has full access to all national police databases in the EU now. It seems as if Washington remote-proxy-controls the EU. Quite amazing to see, too. The amount of bribe money flowing must be legendary.
Surely they are better than Google in this regard.
AFAIK Google is much better at not tearing well-working teams apart in reorgs, and letting people go they have acquihired a few quarters earlier.
Of the few examples I have I my head, I'd even say that the fate of a product that has been acquired by Google is probably even better than those of Google's internally developed products. e.g. Waze is still alive and kicking 13 years after acquisition under its own brand and hasn't been completely swallowed by Google Maps. The Nest brand also stuck around for quite some time.
Google is better but it's an incredibly low bar.
Amazon will drive these people to quit within a couple years over filling out MBRs and threats about how the MBR isn't good enough (MBR = monthly business report, pure bullshit theatre that drives the whole company mad 2/4 weeks every month).
Google would drive them to quit over a longer timeframe, with no threats or harrassment, just because every time they try to ship something cool, someone else has a reason not to do it and that kills you eventually.
They should be fine as long as they remember to use the new cover sheets. I'm sure they got a copy of the memo.
> Amazon will drive these people to quit within a couple years over filling out MBRs and threats about how the MBR isn't good enough (MBR = monthly business report, pure bullshit theatre that drives the whole company mad 2/4 weeks every month).
Wouldn't be possible nowadays to LLM your way out of this?
Really no idea, just asking.
LLM to write it, then all the managers just use LLM to summarise it :)
I'm actually curious too, haven't been there in a while. It's possible that the MBR pyramid has become LLM assisted on all sides but the cultural themes are surely all still there.
Honestly, the whole thing felt LLM to start with, tons of people lacking context and throwing words and summaries around.
> Google would drive them to quit over a longer timeframe, with no threats or harrassment, just because every time they try to ship something cool, someone else has a reason not to do it and that kills you eventually.
Yeah, I've never personally worked at any of those places, but collaborated on a few open source projects with Googlers and Xooglers and the slow grinding wheels of their "processes" that wear people down over time were very apparent. Nevertheless, it has been a breeding ground for many interesting technologies, even if it often suffocates them over the long term.
Why don't they hire MBR specialists and embed them? Leave the development groups alone.
> (MBR = monthly business report, pure bullshit theatre that drives the whole company mad 2/4 weeks every month)
Couldn't have said it any better myself having worked at AWS for close to 8y. I was there during their peak growth years and remember very well how some of these processes that were once not as maddening, devolved.
Google has such a much better reputation when it comes to contributing to and understanding open source. It’s unfair to Google to put Amazon above them.
Heck AWS is the sole reason all these projects needed to go through these license changes to prevent AWS from completely destroying their business models.
Google had a better reputation. But these days I’m really not convinced they’re significantly better than Amazon.
Weirdly I’d put Microsoft above modern-Google, and that’s still a low bar.
..a strategy valid in the good ol' days when a complete rewrite of the currently non-licenseable non-protectable underlying desings and ideas would not cost you 200$ on tokens to reimplement in a different or even in the same language.. maybe SW patents will be a thing soon .. and inadvertently become another nail in the coffin of FOSS and SW development in general, who knows
What business model do MIT licensed projects even have? If you want to offer support contracts and actually make people use them, don't use MIT. You can't have you cake (accepting public contributions and free labor) and eat it (sole profiteering) too.
RMS addressed the business model from the very beginning: it's services and consulting. He practiced what he preached.
You might think that's got a very low ceiling. But, even though it's a bad example in many ways, OpenAI showed that the ceiling is very high. And if you're morally flexible even higher.
RMS promotes GPL, not MIT licensed software.
When a bigger entity (e.g. AWS) decides to undercut the original creator/vendor (e.g. redis,elastic search), MIT code des not help.
RMS would probably agree with you and think that I'm apostate. But there's an open source license for the full spectrum of software authors' preferences. Depending on the circumstances some could turn out to be a poor choice.
I know some lawyers who specialize in these choices. Everyone thinks their choice doesn't smell, but there are the resources to make informed choices.
Actually dual license, those that want industry friendly license can get as much as they are willing to give upstream.
It helps just fine. The original software is still there, available for anyone to use for free. AWS et al can't take that away from us, no matter what they choose to do with their fork.
When they say "doesn't help", they're talking about the developers of the original software you want to use for free getting to eat food and have a roof over their head.
It has an extremely low ceiling. You literally cannot make consumer-targeted software if you want to survive while writing open-source code. Services and consulting is strictly B2B, and services and consulting for open-source software is capable of supporting a population of like 1% of existing software developers. Advocating for this dogmatic ideal that would cull the field by 99% is ridiculous.
I don't know how you came to OpenAI as an example given that they famously succeeded while going closed-source with the release of ChatGPT.
> If you want to offer support contracts and actually make people use them, don't use MIT. You can't have you cake (accepting public contributions and free labor) and eat it (sole profiteering) too.
I can't speak to how successful it is, but Daniel/Curl have been running with this model[0][1] for years. Probably not too different from 'drh and SQLite[2][3].
[0] https://curl.se/support.html
[1] https://rock-solid.curl.dev/
Red Hat grew on GPL licensed projects to become a $34 billion business. Canonical does the same except it’s Ubuntu not RHEL and charges more than Red Hat.
You really underestimate the value of support contracts.
GPL is not MIT. If your software is GPL, you can sell it relicensed as MIT for income.
Doesn’t matter. When companies pay for support contracts they don’t care whether the code is GPL or MIT. They care about the support they get.
I'm not seeing how MIT licensing has anything to do with your ability to use support contacts as a business model. Yeah, nothing constrains the potential customer to purchase support, but that's just as true of GPL licensed software. I can download GPL software and never pay a dime for a support contract, just like MIT software.
Yup, they are very serious about DynamoDB and it is a very good product. Wish they released more information about it, though.
Man I admire AWS marketing team for convincing people DynamoDB is a very good product.
100 items max per Transaction BatchGet 100 items, 16 MB max low write limits on same key Item size 400 KB max etc.
DynamoDB has basically two legitimate use-cases that I'm familiar with:
1. You're selling a system to a customer to use within their own AWS account, that you will have no access to, and it needs a transactional datastore (not just an object bucket) of some kind. The fact that it costs nothing by default (particularly valuable when the customer is trying to deploy a proof-of-concept), scales more-or-less perfectly without anybody touching it, requires zero day-to-day maintenance by you or the customer, and all it will ever ask is that you throw money at it, is very, very much a feature. One example I'm familiar with in the wild is Teleport: https://goteleport.com/docs/reference/deployment/backends/#d...
2. You have a huge OLTP workload that fits Dynamo's KV patterns (e.g. Amazon.com shopping carts, which is what it was originally built for). You don't care how much DynamoDB costs (in either dollars or engineering limitations) because any alternative would melt your face off if you even tried.
Most of the pain that comes from Dynamo is people who try to use it as a primary datastore in place of a relational database just to get the serverless pricing model. It's not worth giving up the flexibility on greenfield systems. It does become worth it to give up the flexibility when your system is mature and you don't have genuine flexibility anymore anyway.
Some of the largest, highest TPS, highest throughput systems in the world use DDB.
The constraints are what let this happen. Unconstraining it might make a better generalist product but part of what you're opting into with DDB is the dumb "put an item in get an item out semantics" and the other side is knowing that it will still work if that volume increases dramatically.
The reason it's good is because, although it is bad, you never have to worry about it in terms of maintenance, admin or scale. It's like lambda for databases.
For the same price, you can run a much more capable PSQL instance with way better features, but now you're on the hook for it being up 24/7.
They should really call it "data-structures as a service" rather than a database, really. Programming against it always reminds me of Leetcode problems, having to define your columns in a way to suit the very specific lookup patterns that they allow for.
I mean this reminds me of Firebase v1 and FoundationDB I think both of which have a favorable reputation for quality in both design and implementation. I do think there's something to be said for being forced to code against what operations are actually fast in your DB. Once it works you can be reasonably confident it will also be fast rather than the typical RDBMS experience of getting an abundance of rope with which to hang yourself with. You often won't find out until it's already load bearing that your query doesn't scale.
Thanks, I don't really get any chance to seriously use or manage OLTP databases (mostly working with OLAP ones myself) so it is good to know about this. The 400KB limit is indeed very limiting -- I had to truncate some data because of that.
I wish I could get a role to work on OLTP databases. PostgreSQL seems to be a fascinating topic so that's on my plate.
They have published 2 academic papers about it. What more do you want, operational details?
For example, I would like to know how to calculate the charge of incremental export before I run one.
One blog says it is charged by the amount of change logs (but the official doc doesn't say so), which makes sense. But how do I estimate the amount? My hunch is: Put + Write + (1~100) * TransitWrite + Update + Delete + (1~25) * BatchWrite.
For anyone interested in the differences between dynamo (the early internal-only KV store described in this paper) and DynamoDB (The AWS service), Marc Brooker has an excellent writeup https://brooker.co.za/blog/2025/08/15/dynamo-dynamodb-dsql.h...
It is their previous one. I did read another paper about DynamoDB, but still there is not much details.
For example, I want to know how to calculate the charge of incremental export. One blog says it is charged by the amount of change logs (but the official doc doesn't say so), which makes sense. But how do I estimate the amount? My hunch is: Put + Write + (1~100) * TransitWrite + Update + Delete + (1~25) * BatchWrite.
> Out of all the big orgs, Amazon is probably the one that has the least regard for keeping technically interesting projects alive
I'm curious what you mean by this. I would have said the exact opposite - AWS tends to keep projects around for a very long time. They haven't acquired very many open-source projects, but the few that they have are all still running as far as I know.
Fig, was one of them but was only partially opens source IIRC. CLI autocomplete and script team sync. Turned into/integrated into Amazon Q, which turned into Kiro CLI, which AFAIK doesn't share any functionality with the original product nowadays.
What good is a foundation when it's core member is owned by a public corporation? Like we know what happens, the pattern isn't unique or rare. Corporate interests will always take priority over community interests, corporations care about profit not community. If Amazon cares about the community as they care about their own workforce, that does not bode well.
who currently has the most regard for keeping technically interesting projects alive?
The title is misleading. AWS acquired DuckLabs, NOT DuckDB. The DuckDB source code is still owned by the nonprofit DuckDB Foundation.
Quote from the article:
"As the CWI representative on the DuckDB Foundation... When DuckLabs spun out of CWI, we created this foundation, which holds all IP of open-source DuckDB, and will continue to do so." - Peter Boncz
You are right.
AWS did not acquire the DuckDB technology itself, which is MIT-licensed open source and governed by the DuckDB Foundation which holds most of the related IP [1].
What they did acquire is DuckLabs, the Amsterdam-based services and development company behind the technology which is owned by and employs the creators and major contributors.
And MotherDuck is a US-based venture-funded commercial company, whose cloud-based data platform is centered around DuckDB but has been significantly expanded recently, including Python pipelines, an agentic context layer, and a visualisation layer.
Interesting to see how they will update their About page.
Currently it says “Today, DuckLabs remains independent and fully owned by the original creators. We've deliberately chosen not to take venture capital, so we can focus on sustainable engineering, correctness, and keeping DuckDB open and MIT-licensed for everyone. “
Future will tell if that ends up being true.
>...sustainable engineering, correctness, and keeping DuckDB open and MIT-licensed for everyone
Hah. Never have I ever heard a sentence that described Amazon less.
what do you mean? amazon maintains a ton of open source projects, including a pure FOSS version of Elasticsearch called Opensearch
Top-level troll: that's not a "version" of Elastic Search, it's a fork of ES specifically created due to ES's licence change that made it more difficult for AWS to extract money from it at Elastic's expense.
Not that I like either actor in this particular play, but this isn't exactly an example of community oriented goodwill on Amazon's part.
No one's asking for "community oriented goodwill", just "sustainable engineering, correctness, and keeping DuckDB open and MIT-licensed for everyone". OpenSearch is a perfectly good example of Amazon maintaining an openly licensed open-source project.
I can’t tell if you are trolling or just oblivious.
Open search wasn’t precisely born out of the kindness of their heart and their defence of FOSS values… it’s more like a predatory practice to capture market.
I would agree that they very much had business decisions around the fork - but calling them malicious is bizarre.
They used the software in a way that complied with the license. When the license changed to no longer suit them for future releases, they forked from the original license.
Remove the names here, and in context, every single person familiar with open source projects and licensing would think this is a nonissue. They used the software as the license was intended, and exercised the rights given to them by that license. That freedom to fork if they don't like where the project is going, or things change and the project is no longer viable for them with the new license, is one of the key freedoms enshrined in the whole FOSS movement.
Amazon is actually pretty good about open sourcing core technologies like that. Firecracker is a big one. They also support Valkey and pay some of its maintainers.
They now updated it. That bit is indeed gone.
Still says
>Independence: We are fully founder-owned and not externally funded. Our priorities align with the long-term health of the project and its users.
This is an important distinction, I hope the title can be changed.
I was really surprised when I first came across DuckDB at how good at it is for its target use cases. It is a game changer for me for the "local analytics" space, and its ability to scale up to a large degree helps a lot.
It is simply awesome to be able to point DuckDB at a mess of CSV and other files and have an instant database on top of it that I can run regular SQL over, and it is fast and just works.
It’s faster than trino and spark for 99% of your use cases. Only once the datasets reach the 100GiB to TiB range is trino/spark faster. Compute is scaling vertically faster than Data Eng/Analytics dataset growth, over time DuckDB will be the better choice.
Not sure why the submitter changed the title from the page: "DuckLabs to Join AWS, Projects to Remain Open Source"
you raised a good point but (et al) didn't touch on the meaningful questions: where does the "core team" of developers of DuckDB now reside? are they now "captured/cashing-in/getting paid" at AWS, or will meaningful MIT-licensed development of DuckDB continue in a robust way outside of commercial enterprises with proprietary needs that are not aligned with where the project was going?
It is not misleading at all. For all these softwares, the team that behind is the most important part rather than the software itself.
Congrats to the founders but honestly feel bad for the team. Hearing so many not good things coming from inside AWS these days. Place sounds like a total mess. Seems like a constant stream of top talent fleeing on my LinkedIn this past year.
I hope they just let the team carry on and don’t contaminate it with all the other craziness going on.
They built something amazing and now have a nice exit. The code remains open under a non profit foundation. I don't think it is that bad. If working conditions are too bad people will just quit and work for the foundation.
An exit can be a nice payout, but only retirement level payouts for a very few. Most will need to find a new job and the market is pretty shit right now and I doubt the foundation has the funds to hire the entire team back.
You make it sound like every worker agreed with this. Is this actually true or is it more about the founders exploiting their workers by selling out to one of the most devious anti-worker tech corporations?
Is profit sharing, like what you're implying, actually happening? If so, in what quantity?
> Place sounds like a total mess.
What big corp is not a total mess?
AWS in large pockets was not a mess 10 years ago.
The things I hear from recent AWS departures are very concerning.
Honestly, there's just a huge amount of thrash across the cloud & SaaS industries as a whole. It's not a lot better at Google (way behind OAI & Anthropic on coding & local AI harnesses), Microsoft (no meaningful frontier model R&D to speak of, and a constantly eroding Windows business), the neoclouds (impossible to source adequate infrastructure and are existentially threatened by the hyperscalers (especially Google) if they can't keep up with demand, chipmakers (NVIDIA vs AMD vs Qualcomm vs Apple vs the long tail of specialty shops like Cerebras), data center pure plays (legislative & community pushback, natural resources, infrastructure availability), and the SaaS folks ("let's see if AI can do it").
It's both a great time to be gold mining in tech, and also a terrible time to be a bit employee.
I started using DuckDB last year and I loved it. Small, focused, crackling with competence, no bullshit. I think what I liked about it had a lot to do with it being made by a small group of technical people who weren't interested in "maximizing share holder value" (that is, maximizing executive promotions and bonuses).
Maybe this will be the rare case where founders survive corporate shenanigans and keep on doing their thing. But I'm bummed, as I've heard that optimism many more times than I've seen it happen.
Seems like a good time to recommend Apache Datafusion. It's designed to be a library, but works pretty well standalone as well. There's a CLI version, but also Python and Java bindings, as well as a Rust library, of course.
From my own experience I can say it integrates far better into your Rust app than DuckDB does.
Well over 100 monthly contributors, too.
Datafusion has all of the individual pieces, especially if you just want to data analysis, but is far from an embedded database.
Indexes, transactions, a first-class storage format are all things that come included with DuckDB, that you won't have with Datafusion.
indexes are very weak in duckdb, need to fit memory.
This fall, DuckDB will support larger than memory indexes! I'm really excited about it. https://duckdb.org/2026/08/17/duckdb-20-highlights#7-storage...
they don't say "this fall", they say later this year, I think it is some parallel effort, and it is actually going for few years already (rewriting index logic to have them in mmapped buffers and not in memory), finger crossed they will finish it finally.
You can even use DuckLake with Datafusion: https://github.com/tobilg/datafusion-ducklake-provider
Is Datafusion related to Polars at all? I ask because you mentioned Rust integration.
No, it's a competing ecosystem, even though both are built on Apache Arrow, and their dataframe APIs might look similar on the surface.
DataFusion is being used as a building block for a growing number of databases and data processing engines in Rust, as it offers the necessary primitives.
Congrats to the founders on their new generational wealth, but also what a shame.
Very much doubt it’s generational wealth. I’m not sure how much people think open source companies are worth, but the path to VC-scale profitability is incredibly rare.
My guess is that Amazon wants official hosted versions and doesn’t want to go through something like the Redis fiasco with licensing. In that case, they ended up having to support their own development anyway (with Valkey), so they might as well just buy the team.
DuckDB team probably gets a nice package and pay bump, but it’s really unlikely they’re getting hundreds of millions from this.
How much do you really need for generational wealth? In my opinion, if you're more than 10 years away from retirement and received a surprise $1 million, you're setup for the rest of your life and your children as well assuming they don't immediately blow it all when you die (that's what a good trust should be for).
Assuming historical returns, your money doubles roughly every 7 years, so within the rest of your lifetime, that 1 million should turn into at least 8. That's an extremely comfortable upper-middle-class lifestyle on the interest payments alone. If your children don't spend it all, your grandchildren would easily have private jet money by the time their parents retire.
1 mil is not generational wealth in the US. It is a big chunk of money, no doubt. It’ll buy a reasonable 2-3 bedroom house in the city I live in, with nothing left over. No one’s definition of generational wealth involves not touching the money for two generations…
The median home price is around $400K. You don't have to live somewhere stupid expensive.
Sure, but that's not generational wealth.
We must be operating on different definitions. Generational wealth means you have enough money to meaningfully improve the lives of your kids and give them a leg-up on life. A million dollars is enough to buy a house for each of your 2.5 kids.
If getting a million dollars wouldn't affect how much money you can leave your kids, you already have generational wealth.
> If getting a million dollars wouldn't affect how much money you can leave your kids, you already have generational wealth.
Well said.
Where I live, one million dollars would allow me to pay off my house, open healthily sized investment accounts for my kids, pad my investment account, setup a trust and, overall, set my family up for a comfortable life in the future. I don't see how that isn't generational wealth.
Generational wealth is where you can also do all of the above for your kids and potentially their grandkids as well.
Basically the bar is higher than "something you can pass down". It is enough that the next generation does not need to worry about making money either.
Generational wealth generally is used to mean something like you and at least your children can live very comfortably off of investment income for the rest of your lives, I.e. none of you have to work for a living. That’s why sibling’s def of 1.5-5m per child is much closer to the commonly understood meaning.
We can debate “live comfortably” if you want, but no 4.5 people are doing that from the investment proceeds of 1m.
What you are describing is kind of more like social mobility.
The commonly accepted definition of generational wealth ranges from around 1.5 to 5 million per child.
I can only find sources that define it the way I said, but I'm sure that folks who think a million dollars isn't enough to buy a house disagree.
Generational wealth is generally used to describe not just a comfortable personal retirement but your children and their children and so on never needing to work if they are remotely responsible with the money.
> If getting a million dollars wouldn't affect how much money you can leave your kids, you already have generational wealth.
Generational wealth is definitely not "affect[ing] how much money you can leave your kids." That's an equivocation - if you're leaving your kids a dollar, another dollar will "affect how much money you can leave your kids."
edit: you need a million dollars to securely retire at all, and that's if your parents, kids, or you don't get sick. If they do, a million is not only not "generational wealth" but it may not even last you three years.
I have to agree, yurishimo's assumption of returns is wildly optimistic.
At a more sane expected return of 5% annually, you get $50k a year to live off of to just keep what you have (or rather, watch it slowly erode in value due to inflation).
That is basically a one person income, maybe two adults if you pinch pennies and live in a crappy apartment or a low-end house in a midwestern suburb.
You can get a lot more lucky with a million bucks than you can with 10k if you gamble, but there are no guarantees. Risk tolerance is the most impactful variable. For those in the low risk tolerance group, I think you'd need at least 2 mil these days. And that's with being frugal, as well as probably not having much left over for kids.
To do that, you would have to never touch the money and invest aggressively. Most wouldn’t do that unless they didn’t need the money, i.e. already rich so not newfound wealth.
You would also have to train your kids to responsibly use the money without demonstrating it, as you’d just be hoarding it. People don’t have a good track record there, either.
I think the best you could do generationally with a million would be to try to invest moderately and draw down a small percentage (2-3%) to demonstrate fully considered use of the money. This would keep you in a middle class income but give you more ability to donate charitably, vacation together, let one spouse retire earlier, solve a financial crisis for a child, etc.
Letting them in on the thinking would give them a good chance to handle a high six/low seven figure inheritance, depending on how the investing goes.
$1M is a newly built condo in the East Bay with HOA >$800/month. Not even an 80 year old 1,300 sqft house in the same area.
Assuming you're investing aggressively and not touching that money for the next 20 years, with the nominal return of 6% per year, adjusted for the inflation rate of 3% per year, 1M will increase to 1.8M. Not sure how you came up with 8M.
3% real return is… pretty pessimistic. You can about get that on long TIPS.
> How much do you really need for generational wealth?
Common definition is somewhere between 1.5 to 5 million per child.
I'd agree with that sentiment but that number would be more like $5M (which is nothing in the context of this acquisition). Inflation is not going to relent and having enough to actually enjoy some of that payout to have pseudo FU money and leave enough invested for children and grandchildren (a well-managed trust).
Buying a house in desirable areas is going to be in the ~ $1M range, college is going to be hundreds of thousands, etc.
Assuming historical returns might be a little unsafe at a point in time near the peak of a bubble popping.
> I’m not sure how much people think open source companies are worth, but the path to VC-scale profitability is incredibly rare.
MotherDuck, a semi-competitor of theirs has raised $100mil in funding. DuckLabs has reportedly taken no external funding (so all ownership is with the founders) and was profitable with 30+ employees.
On the sidelines, other companies behind beloved open source products, like Astral, Astro, Bun are getting bought left and right.
With that as a backdrop, I think they should have been able to get quite a good payout.
DuckLabs is the company that does consulting and DuckDB development, the DuckDB foundation actually owns DuckDB the project.
The DuckDB foundation owns some equity in MotherDuck, which is a data lakehouse platform based on DuckDB, and all three have been moving closely together in making DuckDB better locally as well as in the role of a query engine that really threatens a lot of amazon's role in data lakehouses. DuckDB in the hands of a good team means that you can use AWS almost only for storage, instead of using any of the managed services.
If there's no price tag mentioned, it's likely not generational wealth.
At Google's size, they're not required to disclose acquisition prices well into the hundreds of millions
This is a real shame. I love DuckDB, and we've adopted/extended it and ducklake very heavily at Krea for our research.
I know the projects will remain open, and ostensibly still contributed to in the same direction, but surely AWS thinks of this as another piece of a product suite to compete with Databricks.
Nontheless, I think the open embedded query engine approach DuckDB is spearheading is larger than one project, and I remain excited about the broader ecosystem (especially datafusion).
I'm an idiot, and a lot of Hacker News goes over my head, but I still read it. Would anyone mind answering this question?
Are databases not a solved problem? Why are there lots of different databases? Why is one faster than the other? What's different between them?
It's a fair question!
My attempt at an answer: no, databases are not solved.
Specifically, different databases are better or worse for different use-cases.
For example, Postgres is a great "all around database" - you can use it for a lot of different things. As a "relational" database, it's really good if you have a table full of users, a table full of order, and you want to see all orders made by a user with ID=123. You need to answer questions like that a lot (eg every time someone on a website loads a page) and you need the answer fast (hundreds of miliseconds at most)
However, say your use-case is more like... you've got 100 billion rows of billing data ("joe was charged $123.45 on 2026-03-07 for a shirt, blue, size 11, brand foobar") in one table. You don't care much about joe, but you want to be able to find out how much was billed, total, in 2026-03 for blue shirts (or all year for brand foobar, or all time, for size 11). Postgres would struggle with data of that volume - you'd need a really big expensive database. A "columnar" database like duckdb (or clickhouse) might be able to answer those questions better.
Anyway, different databases are better/worse for:
- Large piles of data that you need to query in seconds
- Huge (petabytes) of data that you need to query in minutes, but can query in parallel
- Many related piles (like a standard relational database)
- Cases where you're mostly getting or retrieving single items (key-value stores)
- Huge piles of data that represent a long stream of events in time (time-series datbases)
- Piles of data that look and act more like files (object stores)
- When you need strict transactions
- When your need is very write-heavy
- When your need is very read-heavy
- and probably many others - I'm not even a huge data guy :)
So it all depends on your use-case. There are still cases that are not served well by any existing database - eg "filtering billions of rows, in milliseconds, by an arbitrary portion of several dozen very-high-cardinality columns" (to use an example that came up recently for me IRL) :)
This is a good summary. It's amazing that new database projects/products (whether it's open source or not) still spring up all the time.
Check this curated list of databases from Carnegie Mellon University: https://dbdb.io/
Just to add to this good explanation, a sufficiently large enough company (and even small successful companies) will eventually have a need for different DBs.
It's all about tradeoffs. Same reason there are so many kinds of wheeled transport and so many kinds of bridges.
There's no single set of requirements and desired properties that people have for databases.
What queries does it accept? How does it persist data? How does it manage replication and partitioning across multiple servers? Are questions with many answers and the right one varies by application.
> What queries does it accept? How does it persist data? How does it manage replication and partitioning across multiple servers? Are questions with many answers and the right one varies by application.
nothing prevents to build single database which would cover all such answers. Its engineering, funding and distribution problems: no-one built it yet.
Practicality and wisdom prevent trying to fill all the database niches with a single product.
Even if you did it, I'd be surprised if a single code base is optimal across the spectrum from resource constrained microcontrollers all the way up to IBM Z Series mainframes and everything in between; along with the full spectrum of persistence from in-memory MRU (or similar) caching to full two phase global consensus as well as optimistic eventual consistency.
There's just too much scope to be the best in class at everything, and even if somehow you did it, the maintenance cost of all the options would be huge.
Best you'll get is as computing continues to increase, the cost of using less than the best becomes more reasonable. There's a lot of database applications where any reasonable database works, and there's a lot of databases that are reasonable in wide application. That may well lead to fewer databases being available, but it's unlikely to converge down to a single database. Just as most engineering domains don't converge down to a single solution for all applications.
> microcontrollers
is a marginal niche, but everything else can be handled by one DB. You think we have PG with extensions (including duckdb) already covers most of the ground, making something like that distributed is also feasible task (there are projects), architecting system to make core engine embeddable is also feasible task.
My very high level take:
Every tool is a trade off between effort to create vs power of the solution.
Effort is generally expensive so most things settle on some general purpose local maximum. If you had infinite effort available, you could build bespoke hardware and software from the ground up to solve every problem. It would be faster and more power efficient than any solution available today.
CPUs win out over integrated circuits because the same CPU can be used for ~every software problem, so by using a CPU you benefit from everyone pooling their efforts to improve the general purpose CPU rather than their own specific niche. But when you reach a certain scale/requirements it makes sense to do something more specific. This is one reason why we have standardized GPUs. Still general purpose but more specialized than a CPU. Or think about how Bitcoin mining moved to ASICs, because they need to do one specific thing as fast and as power efficiently as possible.
So for databases, when you get to specific scale and requirements the same kind of specialization starts to make sense. DuckDB or Clickhouse for analytical loads, TigerBeetle for high scale transactional stuff, etc. And that scale is aggregated across ~all software users, i.e. scale of analytical workloads being big enough to support analytical DBs.
Also as time goes on and industries develop the cost to develop specific solutions can go down.
DuckDB is really great. It's definitely the best tool for working with any type of tabular laptop-sized analytical data. You can do some of what DuckDB does with SQLite, but DuckDB is much more versatile and performant.
There’s trade offs for performance, access patterns, throughput, latency, concurrency, workloads, rigor, type systems, extensibility… really every characteristic you could imagine.
It’s not a solved problem because each iteration of technology doesn’t just fix the mistakes of the past, it’s an evolution to solve the problems of the present.
This is an enormously wide question but the quickest way to give you an idea would be rephrasing as:
"Are cars not a solved problem? Why are there lots of different cars? Why is one faster than the other? What's different between them?"
I think the best way to approach the subject in an easy to grasp way is to ask Gemini or another frontier AI to teach you the basics, they will do a surprisingly good job and they'll be able to react to your questions with INFINITE patience.
> Are databases not a solved problem?
What do you mean by solved problem? I don't own DuckDB (sqlite, postgresql, etc). If I think I can create something as good or better than DuckDB, should I give up doing so (and get filthy rich with an acquisition) because someone thinks databases are solved? Solved databases aren't mine.
I was a real fan from the very beginning, but now DuckDB 2.0 "as a server" and 5. asynchronous I/O [1] looks aimed at something very different from what I thought first.
agree - I was really uneasy with the lede in their v2.0 preview:
"DuckDB has been an in-process database since day one. But people have asked us – very persistently – for a client/server mode, and we have finally caved"
This was already satisfied by numerous projects and products, and feels like a "me too" attempt to capture AI-based workflows. DuckDB always felt like "SQLite for Analytic Data" but I fear these changes and now acquiring the org leading technical direction is where they deviate for good. AWS is so unnecessary for what DuckDB can (and should, IMO) be; MongoDB jumps to mind as a cautionary tale.
Agreed that DuckDB as a server feels far from their initial mission, but how does async IO divert from it?
I didn't even necessarily mean diverting from the mission, just that optimising for network storage now reads like a feature maybe prioritised in anticipation of a stronger cloud user base
Title is "DuckLabs to Join AWS, Projects to Remain Open Source"
This gave me a shot of panic when I read the headline - thanks for correcting the record.
That's horrible news.
What's the relationship between MotherDuck (https://motherduck.com) and DuckLabs?
Hey, MotherDucker here! We've been working closely with the DuckLabs team since day one, and the collaboration will continue! You can read more here: https://motherduck.com/blog/duckdb-amazon/
My understanding is that MotherDuck provides hosting for DuckDB.
And DuckLabs is the engineering behind DuckDB.
If I'm correct, this seems like a signal that AWS is coming after the segment of customers wanting to host DuckDB (MotherDucks customers).
Correct me if I'm wrong though, please.
TL;DR AWS will either acquire MD next/soon, or (more likely) keep them around for the optics that they are totally not going to starve the community of anything other than DuckDB as a hosted AWS service.
From the MD "about page":
The idea for MotherDuck came after Jordan Tigani, MotherDuck co-founder and chief duck-herder, saw DuckDB in action and thought, “Wow, this is amazing! someone should really build a serverless version.”
...
Hannes and Mark, who founded DuckLabs to focus on the core technology and build the world’s best analytics database, were looking for partners who would build a commercial cloud offering.
Maybe they learned AWS were planning on releasing a managed clone without their involvement?
That seems unlikely because AWS needs to build proprietary features within their services to lock in paying (enterprise) users. The problem now is that labs will inevitably focus on this work, which I view as opposed to local, in-process when your paycheque comes from selling hosted services.
Any indication yet as to whether Quack, DuckLake, real-time materialized views dev will continue full-steam ahead? B/c myself and i know of many others very interested in seeing these efforts completed... :crossed-fingers:
1 month later: "Re-writing DuckDB in Rust with Claude Ultra". There is a curios phenomenon, big tech uses great open source projects as a guinea pig for experimenting agent-based development approaches, see Bun, AstroJS.
Wouldn't be the worst thing, when I last used it, I got some c++ specific errors.
AWS rarely acquired startups, though leveraging someone else's tech is a typical pattern. It did license ParAccel and sold it as Redshift. It also runs Athena, which is based on Trino (a Presto fork).
Now, with Snowflake and Databricks earning big bucks in the intelligence era, time to gain market share by having DuckLabs under its ownership?
Also repackaged Blazegraph as Neptune (the original project was shut down after acquisition).
Wow congratulations for the original authors, but I hope it doesn't impact the users. I can imagine that it is very useful for AWS because they don't exactly have something similar.
I use DuckDB extensively for local dev as well as a parquet viewer.
AWS is a great home for DuckLabs. They just want people to use more compute and storage, so they have a degree of technology-neutrality. This is the key to allowing DuckDB to continue to grow in whatever directions are natural, rather than being warped by some kind of walled garden data platform strategy.
Fair point, but no one is clearly articulating what AWS gets out of this. They don't get the DuckDB IP, that's owned by the foundation. All the work that the team at DuckLabs is putting in growing DuckDB and DuckLake are already in principle converting into compute/storage demand in AWS. So what does AWS actually want and how will it change DuckDB to accomplish it?
I made a prediction that motherduck will be acquired by GCP. But I could not imagine duckdb labs will be acquired by AWS. I do not think of AWS as a serious contender in the data space. So, we have now Pola.rs and motherduck in the market. Who will acquire them? I believe Microsoft has brought in a fair few industry talent to build their Azure Fabric product. Then you also have AI companies that are buying devtool/utility companies left and right.
I wasn't expecting this, and I am pleasantly surprised. I think AWS has been a bad player in the data space and very much behind all the other warehouse providers for many years now. This could lead into interesting integrations of DuckDB on top of all the data that lives in S3 and provide a leaner, faster and a more modern alternative to Athena.
Feel like AWS got a lot of the concepts right with S3 Tables* ("serverless" Iceberg catalog on top of S3) ... would love, love to see a version of S3 Tables that lets you choose between Ducklake or Iceberg as the catalog.
* Cloudflare also seems to have gotten the same idea right with R2 Data Catalog
Athena already has Trino and Spark engines, it could totally have DuckDB one too.
What does the financial counterpart of such a deal look like if the IP (main asset) is not part of the deal?
Purely acqui-hiring?
Waiting for QuackDB the OSS spin off any second now
I've already created Haybarn which is Query.Farm's distribution of DuckDB:
learn about it here: https://query.farm/haybarn/
If my extension shows as "build failed" on https://query.farm/haybarn/status/v1.5.5/ how do I go about fixing it?
Presumably the issue is Haybarn-specific?
You can get in touch with me, or open an issue or PR as to why.
The CI extension build runs are still public on GitHub, but may have expired, if you bump your git sha ref it will rerun.
Took a bit of searching around to find the repo: https://github.com/Query-farm-haybarn/haybarn-community-exte...
Looks like this is the failed build https://github.com/Query-farm-haybarn/haybarn-community-exte... but logs have expired
I'll kick it off again for you.
Looks like flaky GitHub this time: https://github.com/Query-farm-haybarn/haybarn-community-exte...
Back in 2018, I took Hannes’ database course as part of my CS degree, which was one of the better ones I took. Over the years, I’ve seen DuckDB celebrated here (and other places), and now this. Congrats to both him and the Ducklabs team.
I've been super impressed by DuckDB ever since first trying it out about 3 three years ago. Congrats to the founders!
What if AWS eventually steer away duckdb team's bandwidth towards their AWS products or influence their work, since they are technically AWS employees.
Current Google cache of this page states:
> "The DuckDB project is governed by the non-profit DuckDB Foundation . The Foundation and DuckLabs are not funded by external investors (e.g., venture capital)."
But "(Last updated: Aug 2026)" and I was not able to find this text anymore.
I wonder if they took vc money after all. I guess not, but still, it seem like to be difficult to live as an independent open source company. Getting a big co as a parent/sponsor is probably the next best thing.
Probably removed because DuckLabs is owned by Amazon now so it’s kind of confusing. Maybe Amazon said, “just take it down, our lawyers will approve some language.”
Congrats to the DuckDB team.
I've been eyeballing DuckDB and LanceDB as part of AI agent memories. This gives me a vibe that AWS will use DuckDB somehow in their ai agents sometime in the near future after seeing the potential.
like vector search ?
Yes. AWS has OpenSearch for cloud-first vector searches (disregarding cloud-hosted postgres with extensions etc). It's ridiculously expensive and not as fast.
DuckDB to the rescue.
Also embed the crap outta DuckDB locally in local agents running on VPS, etc
People using DuckDB, what's your use case? I use sqlite for a lot of my personal projects. What's a good use case for using DuckDB?
DuckDB is for analytics, so OLAP instead of OLTP. If you're using the DB for normal transactions, stick with sqlite
super curious what exactly does it mean to acquire an open source product.
Devs gotta eat. Yeah the code is out there but the code is a result of a lot of thinking, prioritization, research into what is needed from the actual product. So now AWS owns that mindshare and talent and will (probably heavily) influence its decisions. I just hope DuckDB doesn't stray from its goal of being the SQLite (e.g. embedded) OLAP.
When will folks finally learn that if they like having a salary, devs that produce tools actually would like to have one as well?
Never. When it comes to me I am just a poor hustler, surviving a s/w job to make ends meet. When it comes to open source I am gonna hold those guys to highest standard, code, github behavior, tweets, everything must live up to best human ideals.
So give your tools, your life, free bug fixing, priority attention to me because I am getting my paid job done. Why they need money anyway, they can leave on reputation of OSS contributors. Also not to forget I donated 5 dollars last year so now give me full certified audit of your finances of last 5 years.
It can go different ways. Redis, Elastic Search, MySQL, and Java all have different stories with some common threads.
Complete control over what gets merged and worked on by the core team.
it was already like that, and merging 3rd party changes to the core was troublesome, so maybe aws just bought them cause of this reason :D
I am a big fan of the Duck team since before their v1.0 and I am happy for them personally for the big payoff, both in monetary terms but also in terms of scale and longer-term viability of their baby. I am also happy to see the arrangement will keep the core OSS. Let's see what the future holds now!
Remember when big co used to just donate to open source projects to help them out instead of acquiring/dominating them?
Yeah the foundation is doing the heavy lifting here. Amazon's track record with technically interesting projects after a reorg is not great.
I am happy for the team's success but I am sad and anxious for what may become of the DuckDB. It is my favorite tool. Like other comments mentioned, DuckDB is open source and MIT license so I hope it will survive and evolve further the way it did.
DuckLabs does services and support around DuckDB.
"We also worried that scaling DuckLabs into a much larger sales, support, and operations organization would pull our attention away from the technical work and open-source community that made DuckDB successful in the first place."
Given AWS's services arm seems like a good play for a team. Congrats!
Extend, embrace, quack.-
What is the go to alternative from DuckDB?
I'm hoping this means we get an rds + s3 implementation of ducklake from AWS: https://ducklake.select/
did not expect that
I hope the future for DuckDB is still bright
Also did not expect this. I did expect MotherDuck to acquire them at some point (in fact, I kinda thought they already did!)
How do you suppose AWS can monetize it?
AWS acquiring DuckLabs is massive. nice move.
one feature I can't wait from the DuckDb team are real-time materialized views.
I hope AWS doesn’t dilute the DuckDB brand by calling their managed DuckDB offering SageMaker<something>…
I am glad for them. Does anyone know the amount of the operation?
Great move for AWS, does it mean AWS wants to build competitor to clickhouse ?
Shameless plug for DuckDB Harbor - DuckDB server over HTTP (UDS or TCP/IP). Includes a new harbor based DuckDB CLI called pilot:
Avoid antitrust scrutiny with this one weird trick.
Goodbye DuckDB and hello surprise enterprise DuckDB that has special features that you can't live without that only work with Enterprise DuckDB.
We have seen that how many times now? Terraform, docker, elasticsearch, influxdb. Nono, we are still the good guys, believe us. Nothing will change. Yea, thats gonna get forked and rewritten in rust.
AWS is always hungry. Watch out PostgreSQL
This doesn't even make sense. Postgres is and has been open source since the 90s and AWS has had an existing Postgres offering since approximately always.
This will be the engine for redshift now. They’ll make it read write and it’ll be magic.
Prediction: this is start of an aquisition spree by AWS.
Can we stop consolidating everything under a handful of massive companies?
Man. DuckDB is one of my favorite tools in the past decade. It basically replaced pandas/Polars for me and I hope aws leaves it alone.
Otoh, I am glad those guys got paid.
Please edit the title it’s wrong
Congrats!
Insane!!!!
Wow, that sucks balls.
I see a site like https://ducklabs.com/ and it's clear I'm waaaay over thinking my website.
I guess this will be a good test of the open source model DuckDB has opted into.
As I understand it, DuckDB is MIT-licensed and governance is provided by a separate, non-profit DuckDB foundation. This was done explicitly to prevent BSL-ification.
I'm very curious to see whether that approach holds the line. I wouldn't bet on it, but one can hope?
Please No No No why.....
History repeats itself; the same thing will happen as with MySQL—when it was acquired by Oracle, it stagnated.
This is awful news.
such a disappointing outcome
I said "Oh shit !" out loud in a meeting.
Ditto, I had a conversation last week about Duck and Kagi being the two things I hoped never got acquired and here we are.
We just can't have nice things, can we?
Just like press freedom is for those who own press, nice things are for those who can buy nice things.
its open source, we can fork .. (i know for such a project it will require a very special group of "we", but in theory .. it can happen)
If you need a fork, I've already done most of the hard work for all of you.
Or
https://github.com/Query-farm-haybarn
I've been releasing the versions all the way since 1.5.3 and all community extensions.
> a self-contained, signed extension ecosystem
What does this mean? Can I easily sign my own private extensions with my own key?
We are currently allowing unsigned extensions, loading them, then locking the duckdb config to prevent loading unsigned/untrusted extensions
In 2.0 you can sign extensions with your own key and have your own extension repo. Thats the plan at least.
In 1.5 - Haybarn has its own signing key, but its much faster to get extensions released and deployed.
On one hand I am glad that (hopefully) some people were rewarded for their hard work.
On the other hand, do we have to live in an oligarchy?
[dead]
[dead]
[flagged]
I'm glad I stuck with SQLite. Governance is an important part of any open source project. SQLite has kept things narrow and focused, while DuckDB's philosophy is to do more and more, making it a general-purpose data science tool. I am old-fashioned in that I believe in the philosophy that software should "do one thing and do it well"