It’s fine until it’s not… The problem is you can’t really predict when it will fail.

Just make sure you back them up. Bit rot is real.

BASIC? That’s cute.

If you don’t think EE students learn how circuits (including ICs) work, what exactly do you think they’re doing while they’re in school?

OK, but that doesn’t really answer my question, and I’m getting the sense you don’t know how deeply some engineers understand how the hardware works. Plenty of embedded programmers have EE degrees, and can write VHDL just as well (or just as badly) as they can write C and ASM.

You think people writing C(++) for baremetal systems don’t understand how their hardware works?

Even things like HDDs that don’t become “obsolete” in 18-24 months get sold with plenty of life left (unplanned downtime is more expensive than new hardware), but obsolescence makes it happen even sooner.

Regardless of who owns it or what they do with it, those GPUs will get sold on the used market with plenty of life left. Older AI GPUs, networking equipment (eg 100GbE), SAS drives, etc have been easy to find on eBay and other sites for a long time, because data centers replace hardware long before it’s expected to fail.

12gbps could be useful if you use port expanders to put dozens of drives on the same port, but without a port expander you’re right you wouldn’t saturate the 6gbps channel.

This, but if you already have a SAS card in RAID mode you might be able to flash IT (AKA HBA) mode firmware instead of buying a new card.

Also, SAS cables fit SATA drives, but not vice versa. So no need to buy new cables.

Not if it’s for oil based paints.

You seem to be obsessed with optimising one resource at the expense of others. Time is a limited resource, and even if it only takes 5 minutes to configure all of your containers to share a single db backend (it will take longer than that even if you just have 2), you’re only going to save a few MB of RAM. And since RAM costs roughly $2.5/GB (0.25 cents/MB) your time would have to be worth very little for this to be worthwhile.

On the other hand, if you’re doing it to learn more about computers then it might be worthwhile. This is a community of hobbiests, after all…

Neither, I’m trying to explain that you don’t need to know the implementation details of the software running on your server to backup the entire thing.

Where are you getting that from? The fastest and easiest way to back up any server is a full filesystem backup, especially if you’re using something like zfs or btrfs.

I’m saying this based on real world experience: after a certain point you start to see deminishing returns when optimizing a system, and you’re better off focusing your efforts elsewhere. For most applications, customizing containerized services to share databases is far past that point.

Do you have the data to back that up? Have you measured how much of an impact on system load and power consumption having 2 separate DB processes has?

Roughly the same amount of work is being done by the CPU if you split your DBs between 2 servers or just use one. There might be a slight increase in memory usage, but that would only matter in a few niche applications and wouldn’t affect environmental impact.

For most applications the overhead of running a second DB server is negligible.

I write software for a living, and have worked with all 3 database options in the past. I don’t know what DB backend my nextcloud server is using, nor do I care.

They’re obviously basing it on the local laws of the business and customer. That varies from each transaction to the next. They’re just saying that they don’t restrict anything that they aren’t legally required to restrict.

The difference between this and the iPhone 4 leak might just be company policy. I’m sure Apple’s rules for handling prototypes got stricter after the first leak, so this guy probably broke more rules than the first, even if they did basically the same thing.

If you have a chance to read the article, I’d highly recommend it. It directly addresses that point.