I find it somewhat ironic that this seems to have been posted by a literal blogspam bot.

I’m sure that’s totally going to help OCI’s reliability. And Ashburn won’t die for 3 straight days again, like it did this week.

This is such a dumb concept and likely exists entirely to wow investors. A single AI server with 8 GPUs produces something like 7000 watts of heat, if not more. And likewise, will require at least that much power. Sure, solar is “free” once you’ve got the panels in space. The real killer though is dissipating all that heat. Obviously, there’s no atmosphere in space to transfer heat to. Your only option is purely IR radiation which is significantly less efficient. To put this in context, the ISS’s heat dissipation system can dissipate about 14,000 watts. Ignoring all of the other infra that goes into a data-center, that would be two servers. You add all this up, the mass of the supporting infrastructure would far out-weigh the actual servers. And the economies of satellites and rocket launches comes down to mass.

Why are AI agents on the org chart? That’s odd and sketchy. Seems like it could be some sort of fraud to pad numbers.

If you were to actually read the substack the original author wrote, it’s well justified reasoning. The original poverty calculation was based on the cost of food as a percentage of income of a family that is fully participating in society. The author explains though that food is a much smaller portion of our daily expenses and that the cost of fully participating in society includes significantly more expenses. So, if we still use food as a baseline, but re-evaluate it’s percentage of expenses. The new poverty line comes out to about 130k. The author also validates this by looking at the national average expenses and indeed for a family, fully participating in society with no government support, it’s around that range. But you know, continue being snarky.

This is because there isn’t a job shortage. It’s offshoring. The company I (thankfully willingly) left 2 years ago has shifted all of their software hiring to Europe. And since I left has had multiple US focused layoffs. All while the Euro listings keep popping up. And I get it, the cost of living is much lower and the skill set is equivalent. So yea, get your bank. But, this is companies exploiting Europe/Asia, rather than it being something Europe/Asia is immune to.

Yea, it’s the combo of the chiller and cooling tower is analogous to a swamp cooler. The cooling tower provides the evaporative cooling. The difference is that rather than directly cooling the environment around the cooling tower, the chiller allows indirect cooling of the DC via heat exchange. And isolated chiller providing heat exchange is why humidity inside the DC isn’t impacted by the evaporative cooling. And sure, humidity is different between hot and cold isles. That is just a function of temperature and relative humidity. But, no moisture is exchanged into the DC to cool the DC.

Edit: Turns out I’m a bit misinformed. Apparently in dry environments that can deal with the added moisture, DCs are built that indeed use simple direct evaporative cooling.

Practically all even semi-modern DCs are built for servers themselves to be air cooled. The air itself is cooled via a heat exchanger with a separate and isolated chiller and cooling tower. The isolated chiller is essentially the swamp cooler, but it’s isolated from the servers.

There are cases where servers are directly liquid cooled, but it’s mostly just the recent Nvidia GPUs and niche things like high-frequency-trading and crypto ASICs.

All this said… For the longest time I water cooled my home lab’s compute server because I thought it was necessary to reduce noise. But, with proper airflow and a good tower cooler, you can get basically just as quiet. All without the maintenance and risk of water, pumps, tubing, etc.

A coworker of mine built an LLM powered FUSE filesystem as a very tongue-in-check response to the concept of letting AI do everything. It let the LLM generate responses to listing files in directories and reading contents of the files.

Honestly, I don’t mind them adding ads. They’ve got a business to support. But, calling them “quests” and treating them as “rewards” for their users is just so tone-deaf and disingenuous. Likewise, if I’ve boosted even a single server, I shouldn’t see this crap anywhere, let alone on the server I’ve boosted.

In the US, salaried engineers are exempt from overtime pay regulations. He is telling them to work 20 extra hours, with no extra pay.

Commentary from someone quite trusted in the historical gun community and who’s actually shot multiple Welrods/VP9s: https://www.youtube.com/shorts/POubd0SoCQ8

It’s not a VP9. Even at the very start of the video, on the first shot before the shooter even manually cycles the gun, gas is ejected backwards out of the action rather than forward out of the suppressor.

In a centralized management scenario, the central controlling service needs the ability to control everything registered with it. So, if the central controlling service is compromised, it is very likely that everything it controlled is also compromised. There are ways to mitigate this at the application level, like role-based and group-based access controls. But, if the service itself is compromised rather than an individual’s credentials, then the application protections can likely all be bypassed. You can mitigate this a bit by giving each tenant their own deployment of the controlling service, with network isolation between tenants. But, even that is still not fool-proof.

Fundamentally, security is not solved by one golden thing. You need layers of protection. If one layer is compromised, others are hopefully still safe.

If we boil this article down to it’s most basic point, it actually has nothing to do with virtualization. The true issue here is actually centralized infra/application management. The article references two ESXi CVE’s that deal with compromised management interfaces. Imagine a scenario where we avoid virtualization by running Kubernetes on bare metal nodes, and each Pod gets exclusive assignment to a Node. If a threat actor has access to the Kubernetes management interface, and can exploit a vulnerability to access that management interface, it can immediately compromise everything within that Kubernetes cluster. We don’t even need to have a container management platform. Imagine a collection of bare-metal nodes managed by Ansible via Ansible Automation Platform (AAP). If a threat actor has access to AAP and exploit it, it then can compromise everything managed by that AAP instance. This author fundamentally misattributes the issue to virtualization. The issue is centralized management and there are significant benefits to using higher-order centralized management solutions.

deleted by creator

After briefly reading about systemd’s tmpfiles.d, I have to ask why it was used to create home directories in the first place. The documentation I read said it was for volatile files. Is a users home directory considered volatile? Was this something the user set up, or the distro they were using. If the distro, this seems like a lot of ire at someone who really doesn’t deserve it.