You could accomplish what you’re trying by putting the GPU in a second computer. Further, most UPSes have a data interface, so that you could have the GPU computer plugged into the UPS too, but receive the signal when power is out, so it can save its work and shutdown quickly preserving power in the UPS batteries. The only concern there would be the max current output of the UPS in the event of a power outage being able to power both computers for a short time.

Well, my mother has asked me to digitize her collection too and have me host it. Originally, fine, you give your movies to me, I host them, same thing.

Did your mom buy your computer and hard drives? I doubt it. You spent your own money, right? So she’s giving you a whole bunch of stuff which is consuming your space. Quote out the cost of buying components for a separate server for her with her own drives. When she buys the parts, build her her own server and put her stuff on it.

While thats an awesome pile of storage, I weep for your electricity bill.

A bad command execution in large cloud providers can literally make significant portions of the web unavailable, just by the sheer number of services dependent on it.

You can’t have it both ways. You’re trying to call out all of the benefits of running your own infra, but then calling out the downsides of public cloud. Talk apples to apples or oranges to oranges. The point I’m making in the post you’re responding to is that “rolling-your-own” as an organization, specifically a small or medium sized one, comes with risks that far outweigh the costs and risks of public cloud.

The convenience is not worth the risk.

That is not the opinion of non-IT business leaders make decisions to the detriment of the advice of IT departments. You’re ignoring that good IT decisions don’t get to be make by good IT professionals. You’re always limited to the budget and power granted by your organization. That is the practical reality.

So you’re recognizing that a bad command execution can exist in CDN or cloud provider, but where is your recognition of the tens of millions off bad command executions that happen in small IT shops every month?

I looks like you’re ignoring the practical realities that companies rarely ever:

• hire enough support staff
• hire enough skilled staff
• invest in enough redundant infrastructure to survive hardware or connectivity failures
• design applications with resiliency
• have high enough rigor for audit, safe change control, rollback
• shield the operations stupid decisions leads impose because business goals are more important that IT safety

All of these things lead to system impacts and downtime that can only come from running your own datacenters.

The cloud isn’t perfect, but for lots and lots of companies its a much better and cheaper option than “rolling your own”.

How about something less direct like: OffTheCliff.social