Do you need separate systems when you already have Postgres? | Hacker News Reader

Do you need separate systems when you already have Postgres?

postgresisenough.dev

104 pointsb-man86 comments

I love Postgres, and I agree with the general sentiment. But I read the (growing) genre of "use Postgres for everything" articles and they imply a difficulty in running other software that I just don't see.

I'm thinking of Redis in particular. If you're using it as incredibly fast but not critical storage, it's trivial to set up and it ~never crashes or requires maintenance. It creates no headaches, and in exchange gives me a k/v store that I can thrash without worrying about performance (I know it's fast), downstream impact (am I slowing down critical-path SQL queries), etc. Especially in the age of LLMs, which I've found to be great at devops-type tasks, I feel slightly less compelled to simplify my stack.

Redis really is a great piece of software. Low memory, high-performance, feature-rich, and stable. Really hard to beat it for places where you want a durable cache even if technically you can use something else.

Oh no, redis is super simple to set up and use. It's just that setting it up is not as simple as not setting it up.

OK, but it is relatively easy to setup. Obviously, nobody thinks it is _literally_ effortless.

Every extra straw you put on the back of your ops team, especially when you're a young company, adds weight. It's good to be choosey about which technologies actually justify that expenditure.

Your response is a Strawman argument. I was simply saying that nobody thinks that Redis is literally effortless. I was pointing out that the person's comment was not helpful as they were pointing out that Redis installation does actually require effort.

A better response to my comment could have been "just let it go" or something. ₍₍(˶>ᗜ<˶)⁾⁾

For smaller projects and locally, I'm running background services via a compose file... it's a few lines of configuration, and just works. Redis is crazy low hanging fruit.. even job queues that run over Redis make a bit more sense than with Postgres more often than not.

I'm far more hesitant to throw a more formal MQ in the mix though... mostly from experience in that a lot of mid level ("senior") developers don't really understand queues very well at all. Even if conceptually, using tables for queues is more complex.

That said, I will use PG for workflows similar to K/V, Document (JSONB) and other structures over reaching for say MongoDB, etc.

Of course spinning up a hello world of anything is easy. Now introduce monitoring, change controls, version upgrades, multiple environments and regions

It's not that it's necessarily complex. But if you don't need it, don't use it. The business could use your time elsewhere

For a redis cache? I think you're way over-valuing the effort it takes to keep a redis instance running.

For a persistent store.. sure... but that's true for PostgreSQL as well, which has some pretty painful major version migrations by comparison to other options.

Yeah I've found this quite odd. Even the LLMs want to pressure you to not use Redis and go all in on Postgres. Postgres is great, and I usually use it. But Redis is so trivial to add to your stack and it does what it does really well. Why not use the right tool for the job?

because if you already have postgres, and dont need thousands of k/v per second, then postgres can do the same ?

If you're stressing your database, offloading some work to Redis can buy you a lot of Postgres headroom. But sure, start with YAGNI.

LLMs from hyperscalers allow for paid advertising

Presumably this is because LLMs just echo all the pro-Postgres Medium articles they've been trained on. They don't have any actual experience.

And Redis has auto-delete rows (aka TTL). In Postgres you need a cron job to clear stale rows.

This is one way to handle the problem, but not the only...

You can create an index on an expires_at column and opportunistically DELETE on each write to the cache table.

The problem only arises when you have to worry about persistent state. As soon as you have to worry about that, you have to think about backups, replicas, disaster recovery drills and so on. It is much easier to solve for that can of worms for one system than three.

Maybe for the other stuff, but not Redis.

Only problem for me vs using your primary DB is that Redis has no redundancy unless you run it in cluster mode. For us that means when the kubernetes node restarts, availability degrades. Could of course enable clustering, but at that point it isn’t dead simple anymore. And using the DB is.

If you run kubernetes, it might be worth looking at valkey/valkey-operator that manages a valkey cluster for you.

I think your post misses the point of the DBMS centralization: managed consistency.

It is not about ops cost in infrastructure, but ops cost in debugging consistency errors.

I am very much in camp “minimize your dependencies and take the utmost advantage of what you already have” but I also am in camp “your database is the most important bottleneck component.”

This has two implications:

1) make sure if you use Postgres for anything beyond core rdbms functionality that there is no dependency between the two, so you can rip out the additional functionality and move to a different platform when you end up needing to reduce the load on your db server

2) if using Postgres for non-essentials complicates your db backup workflow, risks the data integrity, makes it difficult to maintain or upgrade your Postgres instance (eg you have to wait months or years for compatibility with newer Postgres versions), or loads relatively shoddy or unstable code into the beating heart of your application, then you should either use a different Postgres server/install/container for these ancillary services or bite the bullet and introduce an alternative dependency, depending on which makes more sense.

Basically people can do it in Postgres but at least run separate instances for each purpose. Don't just throw all in single instance.

I'd love the ability to mark a table as "read committed" to prevent long running transactions from keeping old versions of a tuple alive, or even "read uncommitted" to enable in-place updates. Or perhaps instead of downgrading isolation, those serializable/snapshot transactions could simply fail when reading a value from such a table that was modified after they started.

An example of a table that would benefit from this would be rate-limits / concurrency-limits, which are commonly implemented using Redis instead of Postgres.

Surely the last place you'd want to use those weaker consistency guarantees is a concurrency limiter?

The limiter itself should be able to ensure consistency via pessimistic row level locking, even at lower isolation levels.

What I don't want is long running unrelated transactions keeping old tuples alive just it case they're needed. My third suggestion where transactions attempting to read old versions fail is probably better than the first two where isolation gets silently downgraded, since it avoid that problem.

That's interesting, although "read committed" here would be different from the isolation level of the same name. It seems really tricky to implement since tuples could be removed during a query. E.g. the backend could fail to chase tuple-id pointers, and there would be cases where it can be difficult to ensure a tuple is returned exactly once for a given scan.

If you use ordinary READ COMMITTED transactions, they will advance their xmin horizons on each query (and allow old version cleanups) up until their own transaction id but unfortunately not beyond that. For a given table, this is important since that transaction is uncommitted and it might modify the table. If you could make long-running transactions readonly on those specific tables, then you could use a different xmin horizon specifically for those tables. It would require a lot of duplicative bookkeeping in shared memory though. You could probably fake this today by using 2 databases on the same machine and using two-phase-commit+dblink/fdw for cross-database transactions/queries (fdw uses repeatable read in transactions, so it won't allow the xmin to advance).

I'll do you one better: do you really need Postgres and all these extensions when you already have a filesystem?

> only after pushing Postgres to its limits, documenting why it was insufficient, and accepting the operational cost of the alternative

I love Postgres as a DB but, really, this is ridiculous. No doubt these extensions can do the job well-enough but you might as well invest in learning the right tool for the problem from the start, when the stakes are still pretty low. Why wait until, ahem, Postgres is pushed to the limit before you spin up a Redis cluster?

You don't get free opcost by using Postgres for everything. Arguably if you end up with a monolith of a database, you are paying a higher opcost (imagine if too much caching can affect all CRUD ops in your platform). Or you can manage a cluster of PG instances but that's no less complex---each plugin still comes with its own opcost!

No Silver Bullet, No Free Lunch, and all that. If your problem domain really warrants something outside of relational storage, you're gonna pay that complexity cost one way or another. You can't escape it by shoehorning everything in Postgres, fantastic as a DB as it is.

Do filesystems offer ACID guarantees though? Depending on the use case, something like sqlite might work better (or write to a CAS system like S3).

Filesystems aren't efficient for small bits of data.

Like right now I am thinking about a system that has a password reset process and you need to keep track of a user id and a reset token, one or two timestamps, maybe a state variable and a flag or two. That's well under 100 bytes and the cluster size for a typical fs is 4kb or more plus there is the cost of the directory entry. If the OS is Windows it has to ask the Security Manager when ever you open it or delete it which is even more heavyweight.

It's common now for applications that handle lots of little "files" to store them as blobs in SQLlite! See https://sqlite.org/fasterthanfs.html

I can no longer edit my comment above so I'm just leaving this here, picking you arbitrarily out of the handful that addressed my filesystem suggestion:

I am being sarcastic to show what a ridiculous idea it is to shoehorn everything into Postgres. If you find eschewing a relational DB for the filesystem incredulous, you should maybe reconsider if eschewing specialized software in favor of Postgres is advisable.

I don't understand why the answer is to always bloat the system with more specialized software and technical debt, instead of optimizing the existing system. If Postgres can truly handle all of these situations, then mastery of that one tool should be focused on.

I guess its more the rapid start-up mindset to get it up and running fast to sell the company, and leave the problem for someone else which is why a lot of our world is falling apart...

Yeah. I feel it's more of a "ticking all the boxes" situation than anything.

For example: I worked with Rails from 2009-2024 and I haven't come across a single Rails project in the wild that didn't have the queue du jour installed: sideqik+redis, delayed-job before, etc. And then since it's there, people just end up using.

As I don't want to repeat myself, with emphasis added:

> No Silver Bullet, No Free Lunch, and all that. If your problem domain really warrants something outside of relational storage, you're gonna pay that complexity cost one way or another. You can't escape it by shoehorning everything in Postgres, fantastic as a DB as it is.

So as not to bloat the system, you will end up bloating your Postgres installation instead.

Also, Unix principle/KISS: do one thing and only one thing well. Of course this comes back down to your definition of "one thing". "_Any_ Data Storage" seems to be your one thing but not for me. "Relational Data Storage" is a well-scoped one thing for me.

I'm not saying don't take advantage of the plugin system. But we must exercise good sense so as not to abuse it when we avail of its advantages. TFA's suggestion of "only after pushing Postgres to its limits, documenting why it was insufficient, and accepting the operational cost of the alternative" does not sound like good sense. It sounds like a personal and an organizational burnout waiting to happen. Wait until PG is at its limit does not sound like technical debt to you?

> If Postgres can truly handle all of these situations, then mastery of that one tool should be focused on.

Thus you have conceived the conundrum of The Database of Theseus: after how many plugins is your Postgres cluster no longer a mere Postgres cluster to the point where it is unreasonable for a seasoned/certified Postgres DBA to have a good grasp of the system in a reasonable time span?

This is almost textbook https://en.wikipedia.org/wiki/Inner-platform_effect in that this is the second paragraph of the "Examples" section. I acknowledge that Postgres' plugin system mitigates this somewhat but I just can't be convinced that all those plugins will play well with each other for all eternity and especially under stress (unless you make a PG instance for each sub-system, in which case, refer to my original comment).

For the record

- I've used PG for caching with UNLOGGED. Still had Redis for a message queue and other caching. - I've used PostGIS because it made sense to be able to make geo query when latlon is already a (small) part of your schema. - I've used JSONB but only because the alternative is Mongo. I still had an Elasticsearch cluster indexing all that data for search.

In my experience the real answer is almost always "just don't do that other stuff". Most of the complexity Postgress is claiming to fix here doesn't need to exist in the first place.

Filesystem is best when there is a single writer and many readers.

If you have bunch of files and don't have any structure, yes filesystem is great but the moment when you need consistency & performance (which you need sooner rather than later) use databases.

Investing early doesn't hurt when you build a product that you know will have many writers.

For context, I am mainly talking about my personal projects, ideas I am trying out, MVPs/prototypes for potential new businesses, etc.

I have started just using Postgres to back queues. It is simpler (although I have spun up new apps with Redis so many times it is only a small improvement) but more importantly it is cheaper. I really do try a lot of stuff out, so completely removing a infrastructure piece is a nice money saver. Again, not massive but it's cheaper.

The downsides of doing this in the prototype/MVP context are minimal. At the scale of prototype and MVPs, I certainly don't see any difference in performance.

I did note in the graphic it listed Kafka but in the lower table graphic I did not see any Postgres replacement for Kafka. If I am at the scale where I really want Kafka, it is probably for performance and I just can't believe there's anyway Postgres could provide that.

So, I love the flexibility and all-in-one abilities of Postgres for prototyping and making MVPs. But at my job, nobody is proposing exclusively using Postgres for persistence.

I love postgres, but the complexity of using it for everything starts to get pretty high, compared to more tailor-suited tools. We should probably use it for _more_, in general, but the cost of "everything in postgres" is generally higher than I see acknowledged in articles like these.

These specialized tools are likely the cause for the increased complexity within Postgres. I would imagine that if we had more individuals focused solely on Postgres, there would be more discussions, articles, and solutions for a lot of the problems "solved" by adding more tools to the problem.

Articles like these have been insanely effective in biasing LLMs (especially OpenAI) toward Postgres. There areas where Postgres beats MySQL, and vice versa - but ChatGPT/Codex overwhelmingly recommend "default to Postgres."

I agree, the inevitable feedback loops will be painful to watch play out. AI written blog -> AI training set -> repeat.

Until the vibecoders find out they need to manually manage their instances (even if it's RDS or such) and that it all goes over their head

Are the various plugins easier to set up, configure, troubleshoot, etc. than the other software options? If I'm paged awake at 2am, will I have an easier time figuring out pgmq than I will RabbitMQ?

For these specifically? Way easier than rabbitMQ. I love rabbitmq. It's an amazing technology, and if you need a high throughout message bus, it's still a better goto. But.. it does require quite a lot of expertise and know how to operate. Paged at 2am, you will likely get very frustrated. Pgmq though, is "just a bunch of tables". It's going to be a lot easier to figure out what's going on. You already understand transactional DBs.

Unless of course, you already know how to operate rabbitmq.

Two articles on the site are already bookmarks in my app. I completely agree. Not all applications become hugely populair to actually need all those tooling. Plus since I always work in a DevOps team installation and maintenance of all those things are on our team; I don't want to learn to maintain and install all those things. If a team member suggests to add some random tool; the first thing I ask is who will be maintaining it? I'm not.

I have yet to have used UNLOGGED table in production, but I really want to try at some point. Anybody actually replaced a large Redis instance with this?

I have but only to cache data used by cron/bg jobs. We still had Redis for an actual platform system cache. The thing is that Redis cache was created and managed by some framework middleware so it did not seem right to invoke it in code/scripts that bypassed that middleware.

It was ok. Never really failed and didn't affect the user-initiated CRUD ops though ofc the bg jobs ran when the users slept. Still wouldn't bet my life on it.

As small piece of feedback, I think it would be really good if the tools list would include license information, and make it easy to filter for open source tools/licenses. IMHO source-available extensions break part of what is attractive to the Postgres ecosystem, which is that you can move from one provider to another without vendor lock-in.

I was on the Postgres train for 10+ years. Then I fell in love with Sqlite and have been using it for everything.

Fun fact for people whose scaling plan anticipates moving from Postgres to CRDB:

SELECT FOR UPDATE SKIP LOCKED works great for turning Postgres into a job queue. It does not work the same way on CRDB and you will likely have to rewrite those queries or use an external job queue.

> This isn't about dogma. Sometimes you genuinely need specialized infrastructure. But the bar should be high: only after pushing Postgres to its limits, documenting why it was insufficient, and accepting the operational cost of the alternative.

I've seen a few "Use Postgres for Everything!" posts lately. It seems to be fashionable. It reminds me of the Choose Boring Technology[1] thing from 2018 or so, but more specific to a database.

I think the ideas of "don't add unnecessary dependencies" and "ruthlessly evaluate tradeoffs" and "prefer simplicity" and so on are general and have very little to with postgres, so when I see things like "All you need is X" I roll my eyes a little, because these decisions are highly dependent on your use case, and taken as blanket advice it is generally _bad_ advice even if the underlying rationale is sound.

[1]: https://mcfunley.com/choose-boring-technology

ETA: I am going through their list and so much of this means that you are going to manage your own PG cluster and not take advantage of Aurora or RDS, which means you're already committing to a major tradeoff if you want to use a lot of these custom extensions.

I like keeping the number of stateful systems minimal, since those are the most annoying from a reliability perspective. In practice I use one database and object storage (S3). If that database is postgres is secondary.

Couple that with a framework with decent server side rendered template support and htmx for network based interactions and you’ve got 90% of the <acronym_of_the_moment> stack with 10% of the complexity.

> Document store -> FerretDB

Love FerretDB, but it doesn't really replace MongoDB's GridFS which is main reason why most people who are really using Mongo now day. Anyone knows a good replacement for GridFS?

Why would anyone use GridFS for any serious use case? It costs 10-20x as much as storing the files on S3. If unless you already have all your data in MongoDB and don't want an additional dependency for a small number of files it makes sense, but it definitely is not a case for using MongoDB by itself.

Reminds me Kai Lentit's ffmpeg video

https://youtu.be/9kaIXkImCAM

The thing that usually pushes you off Postgres is rarely raw throughput, it's two workloads that want opposite tuning on the same box.

The solution might be to move to separate Postgres boxes everything that needs performance tuning at all.

I'm not in the "use Postgres for everything" camp, but only because I think it's too complex to be used like that. It should be replaced with a bunch of simple primitives in this role. No SQL and query planning magic please.

YES!

The problem of all the datastores is that are applications (like Wordpress) so you are too late to fix anything deep.

We need "frameworks" (so each sub-component can be used as-is or even swapped) and even wondering what a "system level" data engine could be.

agree:

https://github.com/seanwevans/pg_os

https://github.com/seanwevans/pg_git

https://github.com/seanwevans/pg_gpt2

etc...

https://thombrown.blogspot.com/2026/07/load-plcbmbasic81-com...

I suspect a lot of this is resume driven design.

The reason for separate systems is that some of them go down. And by "go down", that includes you wanting to deploy a new version. Or restore from backup as the article suggested.

I think it's rather rare to use different services for redundancy and in most cases if you want that redundancy (e.g. to support red-black or blue-green deployments or just to support better availability via failover) I'd usually suggest using the same technology for that purpose.

I don't believe it's ever easier to manage updating two technologies than one.

I mean if I was only going to use 'one thing', it would be tech that doesn't have downtime when you replace its parts. Something like Kafka or Cassandra. Then since postgres is useful, I would likely end up adding it into the system.

lets take it one step further: Do you really need Postgres when you already have SQLite?

For 90%, maybe 99% of projects: not really.

Postgres is a good middle ground, though.

... and unlike all the new databases, Postgres has a decent license. Everybody else is so afraid of being co-opted by AWS that they won't let you run them the way the way you want.

Is there something specific you wanted to do that was prohibited by a license. I thought most of the licenses you’re talking about just prohibited you from reselling the database as a service.

A client introduced me to Arangodb which I felt was a "secret weapon" that I used for a lot of side projects. Then this came out

https://arango.ai/wp-content/uploads/2025/11/ADB-Community-L...

and it is dead to me. I want my head! I can accept GPL, Apache, MIT or some legit open source license. For my projects I see two possible paths which I want to have open: (1) building a commercial service on top of a database (like my RSS reader) where you can't necessarily draw a clear line between what is allowed and what is not allowed, for instance I have an adaptation layer that makes postgres look like the part of arangodb that I actually use (I do manually rewrite AQL queries into a DSL that extends AlchemyAPI) and if I did something similar over arango is this reselling? (2) an open source project where I want to tell people "go forth and use this code" and not have to hire a lawyer to know what they can and can't do.

Once a vendor has shown they have this attitude, I expect them to change their license for the worse in the future -- I just don't want to invest my time and energy in their platform.

This is why I think sites like this one should show license information and make it simple to filter by OSS license.

Otherwise it's the same trap, just one level deeper.

The license you linked was too much for me because it included limitations 100 GB of data and providing audit logs for the company to inspect.

But I don’t have a problem with people inserting clauses to prevent Amazon from taking over. I don’t expect free work from people forever. If I’m going to use an open source project to build a commercial product, I would only do so if I’m ok forking and maintaining the project myself if necessary.

The license wasn't like that when I first got involved or I wouldn't have gotten involved.

I don't have any fear that Postgres will get relicensed with a worse license than it has. But I see any relicensing or license that is more restrictive than a standard license as a slippery slope that makes me think "I don't want to invest in this platform" thanks to that experience.

I'm a software developer, not a lawyer. I understand standard software licenses and don't feel I have to re-read them or think too much about them. By using one and being dependent on systems that use them I feel like I'm reducing the burden on people who might adopt my open source.

There's a couple of different issues here.

The first is that there are many what most people would call standard licenses that contain tons of restrictions. Many of them like AGPL (and GPL to a lesser extent) aren't even fully understood by lawyers.

Second, when people are talking about standard open source licenses, they are generally talking about licensees approved by the OSI. The problem is that the OSI is primarily financed by tech companies including Amazon, Microsoft, and Google. They're not ever going to add a new category of licenses or promote anything that harms those companies business interests.

As far as relicensing being a slippery slope. There's no such thing as relicensing. Once software is released under a certain license, it is always available under that license.

Anyone can choose to license future work using a different license. That is functionally equivalent to the original authors of a project ceasing work, and someone else starting work on a new fork with a different license. Nothing about Postgres using a standard license prevents this (for what it's worth Postgres is released under its own custom "PostgreSQL License" even if it is very similar to other open source licenses).

What you are actually worried about is that someone stops doing free work, which is the practical consequence of "relicensing". If that's the case you need to look into much more than the license of a project, you need to look into size of the organization, funding, size of the community, time in operation etc... I don't know if operating under a non-standard license is even in any way correlated with an organization or person relicensing or any other form of ceasing development.

It's also postgres, but timescaledb's licensing (and therefore its lack of good support in azure managed postgres) is a bummer.

Yes, that condition makes it no longer open source software.

It also has the effect of making software adopting such licenses getting removed from open source distributions.

If someone takes the MIT license and adds unless your last name ends in ezos then yes it no longer meats the definition of Open Source published by the OSI. But there’s nothing holy about that definition or being “open source”. OSI is just a group funded by companies like Amazon, Microsoft, and Google.

And if they aren’t calling themselves Open Source, then why do you care?

Given the unanimous consensus on HN that Postgres is all you will ever need. I will present 2 resources which I came across recently which changed my perspective on this blind reverence for Postgres for everything.

First is this Oxide and Friends episode [1] where Bryan and gang explains war stories related to operating Postgres during their Joyent days and why they went with Cockroach DB for Oxide.

Second is this amazing blog from brandur which explains several issues with using Postgres as high throughput queue and some mitigations.

Online forums like Hacker news can be a bit echo chamber-y. It is always good to ensure that the people you are taking advice from are solving the same problem as you.

[1] https://oxide-and-friends.transistor.fm/episodes/whither-coc... [2] https://brandur.org/postgres-queues

These are valid push-backs on the "postgres everywhere" argument.

The two critical questions IMO are:

1. At what scale do these issues arise? 2. Does Postgres make it harder to solve them when you reach that scale than another solution would have?

Some of that probably depends on the libraries or tools you use on top of Postgres but being able to easily swap to Redis or a proper queue at scale should mitigate the risk of starting with Postgres if you don't already have the scale.

what about for blob storage too?

Postgres sucks. It does a little bit of everything, but badly and with much manual intervention.

I've come to this conclusion too. It's a bit disappointing. I fully realize this is more than likely just user error. And now with AI agents the maintenance of a Postgres instance is likely less of a burden. But I can't help but yearn for the simplicity of a single SQLite file and bespoke solutions for things like queues, pub/sub, caching, etc.

Wild statement, imo. There is no better open source database, so I'm very curious what you prefer and what your use cases are...

It's boring but the last thing you want is excitement over data that's important to your business!

All the eng. managers I've worked with in the last decade have sworn by Postgres, in the decade before that they were swearing at mongo and getting betrayed by Arangodb switching to a restrictive license and seeing other innovative databases going down the same path means for new side projects I go postgres.

Depends on what you want.

Postgres doesn't do replication, HA or columnar/OLAP workloads, for example.

Reply on news.ycombinator.com