Hacker Newsnew | past | comments | ask | show | jobs | submit | briffle's commentslogin

I hear that argument, but a relative has been an elecrtrician that started out working mostly at the original facebook datacenter in 2016 or so. he now owns the business, and his single biggest client is still the facebook datacenter.

Constant additions, reconfigurations, etc.


It's still contract work. When it's over so is your paycheck.

For a 100MW scale facility the contract work is never over. Once you are done with one bit of work something else is in need of refreshing or changing. Components are breaking daily at that scale, and switch gear, UPS, generators, breakers, etc. all have useful lifetimes and a replacement cycle.

It’s effectively a full time job for an electrician crew or three.

Of course once the facility goes away entirely the job does too. But so goes a factory or anything else.


Construction is one of the jobs that's booming nationwide.

How big is the business?

Should still be orders different from a the continuous labor intensive manufacturing of F35's

Which is a straw man no? This thread is about building data centers, not F35s. Microsoft and FB aren’t competing against LM for land or jobs in Beaver Dam WI nor is it a zero-sum outcome, both can exist ie ‘manufacturing hubs’.

My wife was a surveyor in a past carreer. We have a tape measure in the garage that measures to the nearest hundredth of a foot. It is just so weird to see. Forcing decimal on a measurement that does not normally have it just makes me uncomfortable.

https://surveysupplyinc.com/lufkin-12-foot-hi-viz-engineers-...


The tension that you experienced is common in imperial and traditional systems of units. Here's another example for you: Carpenters working with wood use mixed feet, inches, and fractional inches (like 2' 5 3/8"), but machinists working with metal use exclusively decimal inches (like 29.375"). Both hold steadfast to their traditions and won't consider adopting the other system.

If you take a look at American grocery stores, you'll see things like: “40 × 16.9 FL OZ (1.05 PT) 500 mL BOTTLES / NET 676 FL OZ (5.28 GAL) 20 L” https://www.instacart.com/assets/domains/product-image/file/...

To dissect that product: It's a pack of 40× 500 mL of bottled water. I have zero problems with the metric labeling. But for the US customary units, you can see a jumble: decimal fluid ounces, decimal pints, a large number of fluid ounces, and decimal gallons. Note that the gallons can be broken down into mixed units (and some packaging does that); 676 fl oz = 5 gal 1 qt (0 pt) (0 cup) 4 fl oz exactly. The US units are basically "whatever I feel like using" (don't forget tsp and tbsp, which aren't used in the current example). The metric units for liquid consumer products are always mL and L, which greatly simplifies learning and comparison for the consumer.


Note that ft/100 is almost exactly 1/8", which is also the most standard resolution used in construction. I love decimal feet (it's worth noting that there used to be a survey foot, but it has been deprecated).

I was an archaeologist, but use a similar tape measure. It's fantastic. It also works very well with GIS systems, since decimal degrees almost perfectly correspond to metric powers of 10 at tropical latitudes.

> My wife was a surveyor in a past carreer. We have a tape measure in the garage that measures to the nearest hundredth of a foot

Survey foot or international foot?


> My take is that when Italian customers notice their ping going up by 10x because all their traffic is now routed through France, they will switch to BunnyCDN, Fastly or any of the dozens of CDNs that do have servers in Italy.

While that is true, the datacenters hosting those servers are going to lose a massive amount of monthly income by not having those servers colocated anymore.

And just out of curiosity, how many small/medium websites would have the in house know-how to switch to a different CDN? Cloudflare fronts your site, giving you an 'automatic' CDN, where most others require changes to your site to work with.


The OSU Open Source Lab gives machines to groups in their datacenter: https://osuosl.org/services/hosting/

It has hosted quite a few famous services.


Which famous services?

I doubt OSU is going to host F-Droid. It doesn't even sound like F-Droid would want them to host it.


https://osuosl.org/blog/osl-moving-to-state-data-center/ mentions several major, famous services/projects that OSUOSL either has hosted in the past or is still hosting: kernel.org, Debian, Gentoo, Drupal, OpenWRT, OSGEO. https://osuosl.org/blog/osl-future/ also mentions hosting Mozilla at the time of the Firefox 1.0 release, and having previously hosted Apache Software Foundation. Closer in relevance to F-Droid, OSUOSL hosts the GitLab instance used by postmarketOS: https://postmarketos.org/blog/2024/10/14/gitlab-migration/


F-Droid is the best known non-corporate Android App Store... Why wouldn't they be willing to host it?

It's a critical load-bearing component of FOSS on Android.


They could just license all the IP, and hire away all the Engineers and executives... :)


I don't think that's allowed under the terms of the x86/x86_64 cross-license deal with AMD.

That's why, for example, any meaningful collaboration between Intel and Nvidia under this partnership has to be released in the form of an Intel product using Nvidia tech, rather than an Nvidia product using Intel tech.


Maybe all the engineers, but not the executives who created this problem to begin with


We could just force them into a RAND licensing regime and let the market solve the problem.


no, it wasn't even a real effort. A REAL effort would have been to collect what all is being done, and seeing where things should be removed, processes changed, etc. This was a slash and burn.


Yes because government employees are famously good at judging how to spend taxpayer money.


The idea that they aren’t is nothing more than a right wing meme, presented without evidence while corporate wastefulness is ignored entirely.


[flagged]


In my experience at the DMV they process a large number of clients per hour; they appear quite efficient. The problem is that there aren't enough customer service agents to handle the volume so the lines are massive.

Cutting their budget makes things worse, not better.


Yes: given the number of clients they have and the amount of services they offer, it’s a marvel that the DMV (and USPS) work as well as they do.


Yes. I've been twice in the 9 years that I've been living here. Total time in the DMV in 9 years is under an hour. Last time I went, I spent under 5 minutes inside the building, less than a minute at the desk getting my registration done (I usually do it online, had a weird one-off situation).

I've never experienced customer service half that good from ANY corporation.


Fascinating. Why do you think the stereotype of these hard to fire people is the exact opposite?


I think most stereotypes are the opposite of the truth and it isn't hard to find reasons why.

It's also interesting that you draw a a correlation between "hard to fire" and "incompetent". It's very hard to fire Elon Musk, what does that make you conclude about him?


Noted re: your thoughts on stereotypes. Where do you propose they come from?

Yes, people isolated from consequences do tend to perform badly.

It’s very easy to fire Elon Musk. He stops performing the board turfs him like any other company.


LVM snapshots work well. Used it for years with other database tools.. But make sure you allocate enough write space for the COW.. when the write space fills up, LVM just 'drops' the snapshot.


You'd be even more impressed if you saw just how little resources they have to use (ram, storage, cpu), or how old of a C standard they have to work with. I have a few friends that work on this.


I am indeed impressed but not at all surprised considering what we used to get to the moon!


Seems like Java is popular at Garmin.


And also — sadly — Monkey C. I cannot imagine what possessed them to invent their own scripting language for wearable device apps. It's sort of like JavaScript but worse and with minimal third-party tooling support.

https://developer.garmin.com/connect-iq/monkey-c/


it kinda sucks, but with the constraints it's at least understandable. they wanted an extremely lightweight language with a bytecode VM which could be ported to whatever MCUs in 2015, while also strictly limiting the functionality for battery usage reasons (and, uh, product segmentation/limiting third party access).


These days I'd say "sounds like wasm" but I guess 2015 was a bit before that took off.


There have been enough bytecodes since UNCOL in 1958 to chose from, and embedded is full of them, nothing special about WASM.


Sounds like p-code.


While I might not trust C code more than Java in life saving equipment, I would trust a median C developer over a Java one.


Given the amount of CVEs that would be a bad bet.

High integrity computing is full of pain staking processes, exactly because no one trusts C developers to do the right thing.


Yep, I had a Cyrix processor in mine during that time. Slackware didn't care.


It also worked as a very good space heater.


I had to deal with a state Medicaid system that would go down often. If it crashed after 5pm, it was down until the next morning when someone rebooted the SunOS box. (Yes, they just rebooted the box, and no, in 2014 it was still sunos, not Solaris). Meanwhile, it’s messing up pharmacy authorization for thousand of elderly and low income people in the state ….


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: