Tostino

Heh, yeah I suppose there are still some foot guns if you don't understand how things work.

Using something like EBS for your postgres data directory is, IMO, a bad idea. If you've already made that mistake, yeah I suppose using PG dump isn't that bad.

I'd much rather just do it right though.

Direct attached storage (or whatever storage is fastest / lowest latency for the environment you have available).

Setup pgbackreset or barman, use WAL archiving and setup block incremental backups with a new full backup every so often.

Now you have point in time recovery with very low RPO/RTO, and your database infrastructure can be treated a bit more like cattle instead of a pet, as you should be testing restores often, and this will become part of your yearly major postgres upgrade strategy.

You also get the ability to have a replica for almost free, as replicas can be created from the backup repo very cheaply, and get caught up to the primary without requiring the primary keep around the WAL for a long time. It just pulls it from the repo until caught up.

Any case where you think EBS was the right choice is better handled by having one (or more) replicas and a failover strategy.

Just do the right thing from the start and you save a lot of headaches. People have run production systems already, and have hit the pain points. Why keep hitting the same ones?

But the other option is just doing it right from the start and using a tool like pgbackrest. It's no harder to setup, and it puts you into best practices by default rather than having to work at it later.

I just don't understand why people seem so drawn to the bad solution just because it ships with the database.

Can't recommend pgbackrest enough. It's fantastic software, and I love the work they put into doing inter-file deltas for backups (so if 8kb of a 1gb file changes, you only backup the difference). It saved my last company a ton of money on storage while keeping good RTO/RPO.

It was a decent deal ~6-8 months ago. I had been using 3.1 pro almost since release, but it really is feeling old. After using other models more in the past two months though...I really can't go back to 3.1 pro, as I just have to explain my reasoning so damn much to get it on the right path, where as opus or fable just "get it" from the context of the project much better.

Sonnet is roughly the same level as 3.1 pro for me.

But, they are releasing the weights very shortly (or already have for some of the models discussed). For a very large company, you can purchase or rent the hardware yourself to serve the models.

Or any US hyperscaler with GPUs to spare can decide to serve the models for a reasonable cost/token.

You don't have to send China your data.

At the time the whole thing was led by Yann LeCun

He was not in charge of the Meta LLM stuff, from anything i've read over the past few years.

They were by enough officials that collectively we have voted into power. Maybe we should think about who we're handing power to. They don't seem to have a solid grasp on things.

I've read enough of your comments here to know to disregard your opinion. You just want a surveillance state.

It's a combination of both. Your data layout could be very cachable for one algorithm, but very much not so for another algorithm.

That's just embarrassing. It keeps happening though.

Lol that isn't a database server. Come on, if it has hard drives rather than (good) SSD in 2026, it's not a server you should be using for databases unless it's for something with next to no load, and you are also running your application server on the same machine.

I don't have any personal experience with kratom, other than just recently finding out an old friend is taking it for some reason or another.

Your comment string did remind me a bit about a video that popped up on my YouTube feed like half a year ago from someone I don't watch terribly often.

https://youtu.be/TLObpcBR2yw

Like the other commenter said, Microsoft has zero care for technological supremacy, or even just pushing out a good product. What they do care about is all the useful telemetry data and other signals that get generated inside a browser that they now have access to. Heck, they don't even have to pay to continue supporting the underlying browser technology like they did when they were competing using Internet Explorer.

The license agreement they signed a few years back goes into the 2040s. Long term, you are right obviously. That is just so long term, so much can happen in that time span it's hard to even guess.

Look into Apple's ARM licensing terms. They are very generous to apple.

Decentralization doesn't matter at all if you still have single points of failure everywhere.

Ah, yeah...that's exactly the same thing. Great point. Totally changed my mind!

Edit: Your edit clarifies things a bit. Eh, I got my start in software development working on an emulated Lineage 2 server ~20 years ago, and eventually building my own server from the ground up which extended the game in ways that the normal servers / client didn't support. I think tinkering is fine if that's what interests you.

I don't spend much time these days with games (compared to the past) regardless.

People get busy... I was building my startup around that time period, and didn't really play games for ~5 years.

Why would I reward a company that already stole from me with another purchase?

Who the hell tries something that's been out a few hours and says "My work is not the same"?

Also have a version purchased back when Notch owned. Not played since and have no access. Oh well.

It’s like comparing the experience of buying an iPhone at an Apple Store versus ordering one online.

At the Apple Store, an employee will take off the protective film and actually help you set it up properly (RDS).

Running a benchmark on a default, unconfigured self-hosted Postgres instance is like receiving that online iPhone order, seeing it covered in shipping film, not bothering to transfer from your last phone, and trying to use it anyway.

Concluding that self-hosted is slower based on that is like complaining the iPhone screen is unresponsive while refusing to peel off the plastic just because "that's how it came out of the box." You are expected to take the film off before you use it.

Want to take a stroll on Google maps with me and see all the unused or underused land in the US? People don't actually take up all that much space.

I just did a cross country drive a few days ago. We are not constrained on land.

"wow look at all these options available...to limit users to only use software provided by the same corp" you are missing the point entirely.

I'm sure he could have generated way more total energy if he wasn't trying to get get that max power.

I'm just going to mark you in my mind as a Nazi apologist / sympathizer. You are pushing a real ugly world view there.

I could see something like this working if you actually had a assigned human developer(s) to assist the task. There are few interesting tasks that can actually be completed in one (or few) shot and have anything usable.

Generally fast charging has been a much harder nut to crack than fast discharge. If you have fast charging you necessarily have fast discharge in my experience.

Are you just being intentionally ignorant here? What are you even trying to convey?

It means I'm in the tax bracket that gets to pick up the slack for people who make way more than me but don't pay their fair share.

I want a competent government who directs the funds in an appropriate way. Not one who sets up a shush fund, using my (tax) money, to pay their private army of thugs.