I used MS SQL and Oracle at my last job, but what's there to say about them? They've been around forever, are stable and get all the same table-stakes feature updates as everyone else. Start-ups avoid them like the plague because they're so damn expensive, you won't be running either on your phone or an embedded device like SQLite either.
Is MariaDB included in MySQL? I see no mention of it in the post, but MySQL trending downwards would make sense as people upgrade and switch over. Besides of course novelty wearing off as posited for all engines further down the post
I was wondering the same, but I'm not sure if it would make a major change in the graphs. MySQL and MariaDB have both been unpopular on Hacker News for many years. Submissions on either topic rarely get much traction, which then leads to fewer submissions.
> MySQL trending downwards would make sense as people upgrade and switch over.
No, most large MySQL users are still using MySQL; there hasn't been a widespread migration to MariaDB. They're both actively developed and have grown in slightly different directions. Among corporations, MySQL's usage still far outstrips MariaDB by a significant degree. Lately MariaDB has better product velocity though, and their commercial enterprise finally seems to have stable footing.
> there hasn't been a widespread migration to MariaDB
I don't think I even knew I was running MariaDB at first, or perhaps more as a side note that I saw it dropping in mariadb when I apt installed mysql. If you upgraded Debian some time ago, I'm pretty certain you were automatically migrated, so anyone running that (or, presumably, one of the derivatives like Ubuntu) would have migrated knowingly or unknowingly, hence my assumption
Sure, it's a common point of confusion specifically because a few major Linux distros did that. But SREs / DBAs / DBREs will generally take a much more rigorous approach to database version upgrades. Companies just don't tend to upgrade their important databases in that fashion, and ditto for operating systems if they self-host.
And then there's all the users of managed cloud database offerings (RDS, Cloud SQL, etc) who definitely don't accidentally switch database vendors in that manner. Google Cloud doesn't even offer managed MariaDB, and Azure is retiring their managed MariaDB product.
Also keep in mind MariaDB hasn't been fully drop-in compatible with MySQL for over a decade. They've increasingly diverged in features and minor syntax differences over time.
Just to be clear, I'm not bashing MariaDB, I quite like it as a database. But there's a lot of misconceptions about the relative usage levels of MariaDB vs MySQL among FOSS circles.
MariaDB is widely used, including by some extremely high-traffic sites like Wikipedia [1], as well as some quite large multinational businesses [2].
It may not be as widespread as MySQL, but that's no surprise; despite HN's disdain, MySQL is still one of the most widely-used open source databases in existence.
SELECT
db_name,
sum(if(type = 'comment', 1, 0)) AS comment_mentions,
sum(if(type = 'story', 1, 0)) AS post_mentions,
count(*) AS total_mentions,
sum(score) as total_score
FROM hackernews
ARRAY JOIN
extractAll(replaceAll(LOWER(text), ' ', ''), '(sqlite|postgres|mysql|mongodb|redis|clickhouse|mariadb|oracle|sqlserver|duckdb)') AS db_name
WHERE toYear(time) >= 2022
GROUP BY
db_name
ORDER BY
post_mentions DESC;
Snowflake seems to have peaked; 2023 was hellish dealing with roomfuls of inexperienced devs and even architects convinced it was the fastest cheapest thing ever.
Well as pointed out above since Oracle and SQL Server don't even show up.. this simply does not reflect enterprise and Snowflake and Eatabricks both lean Enterprise
UPDATE: Added a weighted average analysis based on story points and comments. SQLite ranks highest in points per story and Redis ranks highest in comments per post. Also added SQLite to the growth table. I had accidentally deleted this row in the original post.
Interesting to see SQL Server not listed here, am curious whether it didn't have enough signal, or suffered from being a two-word product, with "SQL" being far too generic on its own.
I’ve also don’t remember SAP HANA, Oracle, or DB2 mentioned even once here but believe me, along with MSSQL these occupy most of the top ten database deployments world wide.
Something that I’ve been thinking about a lot recently is that all of the proprietary vendors are quietly strangling their flagship products.
Free and open source database engines were always “nipping at their heels” but weren’t a serious threat for decades. Only other proprietary engines were.
Now that PostgreSQL has more features than SQL Server and better performance, it’s a serious competitor.
But Microsoft is holding MSSQL’s face under water with core-based licensing. It means that per dollar you get dozens of times less compute available for your data than with open-source systems. That ratio is growing exponentially, because they haven’t redone their pricing in… ever.
Oracle and DB2 are being similarly choked off at the same rate, so looking left and right at their direct competition their respective product managers haven’t noticed the problem, which is akin to Fuji and Kodak raising film prices in lockstep just as digital photography is taking off.
We’re entering the era of “kilocores”: single servers becoming available that have over a thousand cores. You can’t imagine what per-core licensing costs for something like that!
PS: I saw a similar dynamic play out in the network space with load balancers and “web accelerators” like NetScaler sold “by bandwidth” with a starter SKU as small as 2 Mbps. I kept trying to politely explain to the reps that the smallest cloud VMs can cheerfully put out 10 Gbps, and hence their product is a 500x decelerator. They eventually listened to someone and made it bandwidth-unlimited. Too late. Everyone uses NGINX now.
Proprietary compilers and developer tooling were similarly strangled, and have been completely replaced by free/open tools in all but a few niche areas such as embedded, hard realtime, and circuit design.
It is also less mentioned on the site in general, owing to it being a proprietary Microsoft product in an audience of people who primarily go for Free / Open Source non-Microsoft products.
There are some people here who are interested in corporate Europe or <insert Microsoft foothold place/industry here>, but most are aligned with Silicon Valley hackers.
Someone else mentioned it already, but what is there to talk about with SQL Server (and Oracle)? Like I'm sure there's plenty someone could write about but generally it's pay Microsoft so it's their problem.
Whereas something like Postgres has a plethora of forks and tools built around it, because it's open source devs can actually do interesting things to solve their problems.
The data query tool linked at the bottom of the post doesn't work for me. Cloudflare shows error 600010, whatever that means. Nice that there is "no login required" but if it did, or allowed that option, maybe it wouldn't need an algorithm to decide whether my traffic is abusive because you could block abusive accounts instead
It does it "very rarely" if you only care about the most populated & richest areas of the world. It also blocks clients from the neglected "global south" all the time. FWIW, I too am stuck in a captcha loop, and these days I usually just bounce when I see Cloudflare captcha instead of trying to fight it. In your logs it probably looks like bot traffic.
Some of the insights match my personal experience and preferences. At $dayjob we're migrating from Mongo to TimescaleDB (now TigerData ¯\_(ツ)_/¯) which is basically a PostgreSQL extension for time series data and couldn't be happier. We are getting better performance and massive storage savings.
On the analytics side of things we are starting to use DuckDB for some development efforts, but we are keen on potentially replacing some or all of our Snowflake usage with DuckDB.
It is the main database for a huge Rails app. They adopted Mongo right when its popularity started to decline. I always thought it was a very poor choice since the day I joined.
It is a especially bad choice considering that a lot of the data stored in it is IoT-like and the system creates a single document per event :facepalm:
I'm sorry to hear about your bad experience. From your comment I take it that you weren't using a time-series collection to store data in mdb which uses industry-standard compression techniques?
Still Public, still chews through million->billion or rows in seconds. Their Cloud version has some Cloud specific features. A few vendors have build custom thing on top or custom builds off the open source project too.
If your curiosity inspires you to dig deeper into the data, our "chat with hacker news" free tool is available. No login required: https://camelai.com/hackernews/
the funniest thing about this graph is that it proves there was a raw drop off in all popularities in the last 2 years, which of course directly coincides with the great layoffening that has been happening for almost 3 years now.
this shows that people are definitely rotating out of "web technologies" in general, not because they aren't useful, but because the money isn't there anymore.
perhaps a large chunk have switched to AI hype trains, and it would be interesting to compare raw results of different AI headlines, but i suspect maybe 30% of people have left tech all together.
This would be an interesting request to directly ask the data. Which you can do using our "chat with hacker news data" free tool: https://camelai.com/hackernews/
i went there on mobile and asked two questions. it went pretty well from a UI and response quality perspective. the data they showed me didn't show any obvious trends, but i suspect it's because i didn't specify a long enough list of technologies, and that some general terms were included like "machine learning" and "llm" which had an effect of hiding the trends i was looking for.
a great start and much more enjoyable than writing the sql or for loops myself :)
Absolute drivel. Comparing operational/transactional databases like MongoDB and Postgres to analytics / columnar datastores like Redshift and Snowflake is meaningless. You might as well as say "...the popularity of hammers is way up, with screwdrivers appearing to be in decline..". If this is the type of data analysis that AI is supporting, we're all in trouble.
MS Sql Server not even mentioned. This tells us there is a whole world almost totally omitted from discussion on HN: "Enterprise"
Oracle isn't in there either, which goes to show how much of a bubble HN actually is considering MSSQL and Oracle are #1 and #2 in market share.
I used MS SQL and Oracle at my last job, but what's there to say about them? They've been around forever, are stable and get all the same table-stakes feature updates as everyone else. Start-ups avoid them like the plague because they're so damn expensive, you won't be running either on your phone or an embedded device like SQLite either.
There is a reason it is not even mentioned
More unsolicited feedback: Month-by-month is kind of noisy. You might do 3 month average to smooth it a little and make the trend clearer.
Is MariaDB included in MySQL? I see no mention of it in the post, but MySQL trending downwards would make sense as people upgrade and switch over. Besides of course novelty wearing off as posited for all engines further down the post
> Is MariaDB included in MySQL?
I was wondering the same, but I'm not sure if it would make a major change in the graphs. MySQL and MariaDB have both been unpopular on Hacker News for many years. Submissions on either topic rarely get much traction, which then leads to fewer submissions.
> MySQL trending downwards would make sense as people upgrade and switch over.
No, most large MySQL users are still using MySQL; there hasn't been a widespread migration to MariaDB. They're both actively developed and have grown in slightly different directions. Among corporations, MySQL's usage still far outstrips MariaDB by a significant degree. Lately MariaDB has better product velocity though, and their commercial enterprise finally seems to have stable footing.
> there hasn't been a widespread migration to MariaDB
I don't think I even knew I was running MariaDB at first, or perhaps more as a side note that I saw it dropping in mariadb when I apt installed mysql. If you upgraded Debian some time ago, I'm pretty certain you were automatically migrated, so anyone running that (or, presumably, one of the derivatives like Ubuntu) would have migrated knowingly or unknowingly, hence my assumption
Sure, it's a common point of confusion specifically because a few major Linux distros did that. But SREs / DBAs / DBREs will generally take a much more rigorous approach to database version upgrades. Companies just don't tend to upgrade their important databases in that fashion, and ditto for operating systems if they self-host.
And then there's all the users of managed cloud database offerings (RDS, Cloud SQL, etc) who definitely don't accidentally switch database vendors in that manner. Google Cloud doesn't even offer managed MariaDB, and Azure is retiring their managed MariaDB product.
Also keep in mind MariaDB hasn't been fully drop-in compatible with MySQL for over a decade. They've increasingly diverged in features and minor syntax differences over time.
Just to be clear, I'm not bashing MariaDB, I quite like it as a database. But there's a lot of misconceptions about the relative usage levels of MariaDB vs MySQL among FOSS circles.
is anyone seriously using it? even their own brand facepile is pretty weak
MariaDB is widely used, including by some extremely high-traffic sites like Wikipedia [1], as well as some quite large multinational businesses [2].
It may not be as widespread as MySQL, but that's no surprise; despite HN's disdain, MySQL is still one of the most widely-used open source databases in existence.
[1] https://wikitech.wikimedia.org/wiki/MariaDB
[2] https://mariadb.com/resources/customer-stories/
Their what now?
There's an online playground with the data here: https://play.clickhouse.com/
Wrote up this query:
Sqlite seems to be growing recently which matches my perception, but it‘s not listed among the growing databases. Weird.
Yeah I found a mistake in the analysis. I'm updating the post to reflect SQLite's popularity.
SQLite is now reflected in the growth table
Snowflake seems to have peaked; 2023 was hellish dealing with roomfuls of inexperienced devs and even architects convinced it was the fastest cheapest thing ever.
Well as pointed out above since Oracle and SQL Server don't even show up.. this simply does not reflect enterprise and Snowflake and Eatabricks both lean Enterprise
UPDATE: Added a weighted average analysis based on story points and comments. SQLite ranks highest in points per story and Redis ranks highest in comments per post. Also added SQLite to the growth table. I had accidentally deleted this row in the original post.
Interesting to see SQL Server not listed here, am curious whether it didn't have enough signal, or suffered from being a two-word product, with "SQL" being far too generic on its own.
I’ve also don’t remember SAP HANA, Oracle, or DB2 mentioned even once here but believe me, along with MSSQL these occupy most of the top ten database deployments world wide.
Something that I’ve been thinking about a lot recently is that all of the proprietary vendors are quietly strangling their flagship products.
Free and open source database engines were always “nipping at their heels” but weren’t a serious threat for decades. Only other proprietary engines were.
Now that PostgreSQL has more features than SQL Server and better performance, it’s a serious competitor.
But Microsoft is holding MSSQL’s face under water with core-based licensing. It means that per dollar you get dozens of times less compute available for your data than with open-source systems. That ratio is growing exponentially, because they haven’t redone their pricing in… ever.
Oracle and DB2 are being similarly choked off at the same rate, so looking left and right at their direct competition their respective product managers haven’t noticed the problem, which is akin to Fuji and Kodak raising film prices in lockstep just as digital photography is taking off.
We’re entering the era of “kilocores”: single servers becoming available that have over a thousand cores. You can’t imagine what per-core licensing costs for something like that!
PS: I saw a similar dynamic play out in the network space with load balancers and “web accelerators” like NetScaler sold “by bandwidth” with a starter SKU as small as 2 Mbps. I kept trying to politely explain to the reps that the smallest cloud VMs can cheerfully put out 10 Gbps, and hence their product is a 500x decelerator. They eventually listened to someone and made it bandwidth-unlimited. Too late. Everyone uses NGINX now.
When you're addicted to bad revenue is very hard to compress it
It's a repeating problem across many industries.
Proprietary compilers and developer tooling were similarly strangled, and have been completely replaced by free/open tools in all but a few niche areas such as embedded, hard realtime, and circuit design.
It is also less mentioned on the site in general, owing to it being a proprietary Microsoft product in an audience of people who primarily go for Free / Open Source non-Microsoft products.
There are some people here who are interested in corporate Europe or <insert Microsoft foothold place/industry here>, but most are aligned with Silicon Valley hackers.
Someone else mentioned it already, but what is there to talk about with SQL Server (and Oracle)? Like I'm sure there's plenty someone could write about but generally it's pay Microsoft so it's their problem.
Whereas something like Postgres has a plethora of forks and tools built around it, because it's open source devs can actually do interesting things to solve their problems.
Unfortunately, I only got data until 2022, but here is a similar overview with a few more charts and sentiment analysis: https://eu.getdot.ai/share/f3f0853d-fa91-4301-8fb2-52821b65e...
Will try to update it with some more recent data later.
The data query tool linked at the bottom of the post doesn't work for me. Cloudflare shows error 600010, whatever that means. Nice that there is "no login required" but if it did, or allowed that option, maybe it wouldn't need an algorithm to decide whether my traffic is abusive because you could block abusive accounts instead
https://camelai.com/hackernews/? Worked for me.
I am getting an infinite loop of 'Verify you are human'....
We use cloudflare turnstile. Sometimes it blocks some VPNs. Very rarely it blocks some browsers.
It does it "very rarely" if you only care about the most populated & richest areas of the world. It also blocks clients from the neglected "global south" all the time. FWIW, I too am stuck in a captcha loop, and these days I usually just bounce when I see Cloudflare captcha instead of trying to fight it. In your logs it probably looks like bot traffic.
Not on a VPN. Guess I can't use this browser then? So much for HTML/Ecmascript being standards anyone could implement...
Yep, that one. Praise that the algorithm likes you!
Some of the insights match my personal experience and preferences. At $dayjob we're migrating from Mongo to TimescaleDB (now TigerData ¯\_(ツ)_/¯) which is basically a PostgreSQL extension for time series data and couldn't be happier. We are getting better performance and massive storage savings.
On the analytics side of things we are starting to use DuckDB for some development efforts, but we are keen on potentially replacing some or all of our Snowflake usage with DuckDB.
Can you tell me, the scenarios you used MongoDB for? Because I'm still curious about why would anyone use MongoDB after all these years.
It is the main database for a huge Rails app. They adopted Mongo right when its popularity started to decline. I always thought it was a very poor choice since the day I joined.
It is a especially bad choice considering that a lot of the data stored in it is IoT-like and the system creates a single document per event :facepalm:
I'm sorry to hear about your bad experience. From your comment I take it that you weren't using a time-series collection to store data in mdb which uses industry-standard compression techniques?
>a ClickHouse database of every HN story
I remember downloading it a few years ago, but the bookmark I have is dead. where is it now? is it still public?
Here: https://play.clickhouse.com/play?user=play#U0VMRUNUIG1heCh0a...
It's really fantastic. Continuously updated and fast anonymous queries. Big kudos to ClickHouse.
Still Public, still chews through million->billion or rows in seconds. Their Cloud version has some Cloud specific features. A few vendors have build custom thing on top or custom builds off the open source project too.
How are you handling sanitization? Anything interesting?
Would be great to share the queries. Are these results weighted for storypoints and/or number of comments?
Purely based on headline occurrence but weighing based on storypoints and comments is a great idea. I'll update the blog, thanks.
Updated with weighted analysis.
If your curiosity inspires you to dig deeper into the data, our "chat with hacker news" free tool is available. No login required: https://camelai.com/hackernews/
the funniest thing about this graph is that it proves there was a raw drop off in all popularities in the last 2 years, which of course directly coincides with the great layoffening that has been happening for almost 3 years now.
this shows that people are definitely rotating out of "web technologies" in general, not because they aren't useful, but because the money isn't there anymore.
perhaps a large chunk have switched to AI hype trains, and it would be interesting to compare raw results of different AI headlines, but i suspect maybe 30% of people have left tech all together.
I think it's attention and mindshare going to AI
we would have to look at raw numbers, like, perhaps web tech is just "flat", not declining.
but my suspicion without evidence is that the gross number of people in the industry is actually dropping, though it should be increasing.
This would be an interesting request to directly ask the data. Which you can do using our "chat with hacker news data" free tool: https://camelai.com/hackernews/
No login required.
i went there on mobile and asked two questions. it went pretty well from a UI and response quality perspective. the data they showed me didn't show any obvious trends, but i suspect it's because i didn't specify a long enough list of technologies, and that some general terms were included like "machine learning" and "llm" which had an effect of hiding the trends i was looking for.
a great start and much more enjoyable than writing the sql or for loops myself :)
thank you very much for suggesting that and for making it available without a login :)
Confusingly, I just came across the unrelated https://www.camel-ai.org/ today.
Sooo confusing. We've debated changing our name but can't bring ourselves to break up with our cute camel logo lol.
I would have suggested HumpAI if hump didn't have another meaning that might attract users you didn’t intend :)
XD lmao
First one to http://camel.ai/ wins :)
I had a semi-viral tweet about my attempt to buy camel.ai useless domain squatter wants $40k for it! https://x.com/isabella_patane/status/1820987472287080867
No SQLite?
Mistake in the analysis. Fixing now.
Fixed.
[dead]
Absolute drivel. Comparing operational/transactional databases like MongoDB and Postgres to analytics / columnar datastores like Redshift and Snowflake is meaningless. You might as well as say "...the popularity of hammers is way up, with screwdrivers appearing to be in decline..". If this is the type of data analysis that AI is supporting, we're all in trouble.