cduzz

I think it's a matter of scale and focus.

Dell is at a sweet spot for many organizations where you can get hardware support from dell, or not.

You can (with dell) just buy used junk and use that, and you can still access their support library for firmware updates.

I did a lot of work with storage with dell stuff and I found it far simpler to "outsource" the storage integration to dell than to do it myself -- they make sure that all "2tb disks" are the same (exact!) size, they all use the same mechanism for firmware updates, etc.

A much larger org could have saved some money doing that integration and testing internally; a much smaller organization could save money just buying stuff from wherever and not worrying.

I personally haven't used dell support in over a decade, but I have found that their "look up the serial number" web site has been invaluable; I've found the fact that they publish firmware for all their stuff and let anyone install it valuable; the huge secondary market for dell is helpful, etc. And, if you care, you can even give them more money to directly support you, instead of just doing it yourself.

HPE, on the other hand, locks their stuff down super hard; I assume there are reasons to go with HPE, but I'd rather get clocked on my superdome with an HPUXy stick than ever touch that stuff again.

I was fortunate enough to work with a fair bit of sun gear (3/260, 4/670, enterprise 250/450 and even some big iron 6500s) back in the day.

These days, I've got several Dell 5820 style workstations and I'm pretty sure someone who interned at sun in the 1990s is now running dell's workstation division. The medium sized dell deskside towers have nearly identical vibes as those old workstations, and are a pleasure to work with.

I felt old when I realized that someone had put my first "big boy" unix computer, a decstation 3100 with 24 mb of ram, onto a business card.

https://dmitry.gr/?r=05.Projects&proj=33.%20LinuxCard

I did systems administration for a university math department for several years. I came to the conclusion that mathematics (and perhaps philosophy) were both topics where it was likely that no staff members in that department could describe "what goes on here" and that possibly even within the department, one professor may not be able to describe what another professor's actually doing.

The closest I could come to describing math is "some abstract process where imagined structures are characterized and extended; the most critical part of the process is identifying where seemingly independent structures are found to actually be fungible in some previously undiscovered way".

A simple example is

  "hey, did you know that x^i is the unit circle?" 

  "what's i?" 

  "i is defined as if you square it the result is -1" 

  "what does that have to do with circles?"

Fun bugs in missiles are sometimes unfun.

https://www-users.cse.umn.edu/~arnold/disasters/patriot.html

That sounds like "MII" -- the 100mb version of AUI?

And also when deleting a file that's on NFS, but it's reference count is non-zero in the kernel (because at least one process has it open) you end up renaming not deleting the file.

I'd guess that any of these media is sketchy for long-term archival storage.

I've got probably 30 5.25 floppies in my house, and I have a 5.25 drive and an old motherboard that can probably recognize that drive, and I've got the cables to connect the drive to the motherboad (and power supply). I'd guess that there's maybe a 20% chance that, if I spent a week trying to mash it all together I'd be able to get some data off of some of these floppies.

On other magnetic media the chances are between "haha, no" (oh, you've got an old 8 inch 10mb Bernoulli disk?) to 5% (oh, a QIC150 tape).

On the other hand, if I found a stack of moldy greenbar printout of some data, I'd say that most of the data can likely be recovered, and likely for not a tremendous amount of money.

Would they announce record losses and financial distress?

We'll see when they go public. Until then all these press releases are strategic messaging...

Zitron's claim is that much of that boost in profit for microsoft and alphabet and amazon is from only a few customers, specifically openAI and anthropic, and that those companies are buying and promising to buy lots of resources with money from investments from those same companies, and that largely this is unsustainable unless openai and anthropic can find a profitable business model.

I have watched many of his screeds; I think he undercuts his thesis with excess bile.

His core observation is that the unit economics of openAI and anthropic don't actually yield enough profit to pay off the huge debts that these two companies have incurred, and that as a result all the debt they've taken on will have to be written off which will trigger "the hyperscalers" to themselves suffer huge losses (likely wounding google and microsoft and destroying oracle).

He is rather more negative about the utility of LLMs than lots of other people (myself included); but his overall view of

  1. they're not completely trustworthy
  2. they're really expensive to train
  3. it isn't obvious that anyone's willing to pay the full freight for the resulting product
  4. lots of large orgs "that should know better" have gone far down the LLM "AI" road because they're looking for "the next big thing" when they should be pivoting to "mature, stable" companies instead of "hypergrowth" companies.
  5. there's lots of debt and obligations and no obvious way for all of it to be paid off from revenues from openai / anthropic.

seems pretty reasonable. There seem to have been lots of bets placed on the hope that this stuff will continue scaling as it has in the past, if it is given more compute and data, and that bet is one that has yet to have demonstrated itself as correct.

Elsewhere, people have pointed out that many of the "FAANG" companies have shed lots of people and driven lots of profits, largely on the back of internal use of LLM tools. That doesn't necessarily contradict the skepticism that anthropic and openAI will succeed, and given all their obligations, if they fail it'll be a big mess.

But again, big Z doesn't do himself any favors when he rants about "failsons" or whatever.

Sure, that's possible?

It's also possible that, because the US is busy bombing Iran, Iran may be busy attacking the US infrastructure in any way possible? The US is also in a tepid war with russia, last I checked.

Some "cyber" crew getting a shell on an outsourced service provider and running a "defrost" command is also a totally explanation for this situation. It's also totally possible that some crew has mapped out a list of PLC entry points for various orgs and has them in a spreadsheet of "if we find a vulnerability, we should X this Y with this prestaged script that our intern / LMM cooked up last year to defrost these freezers"

I wonder if you could run kafka on k8s on VMS on VMs on ec2?

https://aws.amazon.com/blogs/migration-and-modernization/dep...

The US Constitution, specifically the 13th amendment, allows:

  Neither slavery nor involuntary servitude, except as a punishment for crime whereof the party shall be duly convicted, shall exist within the United STates or any place subject to their jurisdiction.

So in fact, forced labor camps are totally legal and common in the United States. There are lots of private companies which operate these prison / factory systems and are quite profitable (unsurprisingly), such as NYSE:GEO.

All governments "redistribute wealth" friend, not just the left or right or up or down.

It all depends on which side of the bars you live on, amirite?

That reminds me of core memory, which is also nonvolatile and for a long time was the primary memory of many computer systems.

I suspect "growing up" with a system where the computer itself is nonvolatile (turn it off, move it, turn it on, it's doing exactly what it was doing when it was turned off) changed the way people looked at computers vs modern systems.

It's too bad intel didn't push harder on their optane stuff. I guess it's too bad (for intel) for lots of different abandoned / failed projects.

LLMs are golden retrievers that have memorized all of wikipedia

Just to help me out a little; you're saying "printing press and electricity"

But there were printing presses before movable type, and movable type specifically allowed creation of inexpensive books and newspapers. The printing press itself wasn't all that useful alone.

Electricity (in the form of shocking fish) was used by people thousands of years ago and Volta was fiddling with electricity to make frog legs dance in the 1800s; electricity as an industrial property required a bunch of knowledge and material science to make ... lights and motors. Electricity by itself isn't all that wonderful; a flash in the pan so to speak.

So -- what "electric light" or "movable type" is the product of LLMs? I'm sure there are, but ELI5...

I can think of "Summarize this thing for me" and "Elaborate on this half baked idea" -- and honestly "summarize this for me" is actually quite useful. But I'm not sure it is "electric lights / electric appliances" or "indoor plumbing" degree of revolutionary.

I am planning on fiddling with using the TPM of an old dell 5820 I've got floating around my house as a budget HSM.

It looks possible, on paper, if you don't poke at it too often.

And I would have put the bearing shell over a pvc pipe and used a hand drill or just spun the drill bit between my fingers, assuming it is just bearing material.

Those kids (the autopian team) specialize in generating engagement doing things that are pretty unsafe...

Another example would be the dude who used a chainsaw on some lead acid batteries, for reasons.

https://www.theautopian.com/how-i-used-a-chainsaw-to-remove-...

That dude's also an owner of the company, so maybe giving himself lead poisoning for clicks is worth it?

DRAM is highly regular and relies on a 1T1C (one transistor, one capacitor) structure. Because those tiny capacitors naturally leak charge, the adjacent transistors must be optimized for ultra-low leakage. That requires thicker gate oxides and higher threshold voltages which makes them slow. Furthermore, digging those deep trench or stacked vertical capacitors requires highly specialized, high-aspect-ratio etching and deposition tools. Also, for DRAM the backend wiring is completely different; DRAM is a giant, uniform grid that only needs about 3–5 metal layers to keep costs rock-bottom while modern logic is highly irregular and complex, requiring 12–16+ metal layers.

Logic processes (CPUs/GPUs), on the other hand, are optimized for raw switching speed. They use thin gate oxides and low threshold voltages, which leak like a sieve. This is fine for a fast processor, but it would wipe out a DRAM cell's charge almost instantly.

Cisco has trained a generation of home-lab would-be cisco engineers to dig through shady russian forums to download sketchball binaries. I think they even hide the checksums of their gated binaries, just to make sure you're never sure how sketchy their code might be.

But in the one specific case of mobility express, you can ... just download it. Unexpected but actually pretty cool.

You still have to figure out installing and configuring it, which is mildly tedious, and you'll probably need one of those cisco serial cables, but it is an exciting side-quest instead of a dirty forum crawl.

To bring it back to "openwrt" -- this nonsense by cisco, broadcom, HP, IBM, etc is why generation after generation the "enterprise" market is dying -- they're excluding 90% of the would-be engineers from the job market protecting their e-waste from secondary or later reuse, likely just on the off-chance they might make a tiny amount of money downstream. But I guess that's not a line-item that shows up in the quarterly reports (what's the value of "community goodwill" ?)

I'm overall not a fan of Cisco; quite the opposite. There are lots of companies ahead of Cisco in the "pretty not so cool" these days so my ire is not as strong as it was when I was a young idealist.

As far as I can tell, in this one case (with mobility express) it seems like Cisco accidentally did something overall (IMO) "good" for the world by making it possible for their kit to be reused by some part of the world where someone will get utility out of the still very nice equipment and somehow cisco won't see a dime from that "someone is using cisco equipment" event.

Mobility express isn't perfect, but considering the cost of the gear and the cost of the license, it is pretty darned good. I suspect some product manager at cisco got fired for accidentally making the world not quite as bad a place as it could be if cisco extracted license fees on all the 2802, 3802, and 4800 access points or otherwise sent them to the e-waste bin 4 years early.

Thanks for the head's up! The Rukus Unleashed looks like the perfect replacement for my icky system.

I'll put in an ebay search notification for when the R650 (and R750) for $50 each and maybe it'll ding in a couple years and I'll be in a place to swap out the 3802 network I've got running now...

I didn't look at those; do they support running some controller thing in one of the APs to allow central management of all the nodes?

Cisco's mobility express just runs on one of the APs and can fail over to another of the APs; it's a slick piece of software.

And yes, it isn't open source, which is a real shame since cisco's killed it (as far as I can tell) and it probably represents an enormous and sophisticated investment in effort and engineering and it'll just melt into entropy.

I loath cisco and don't recommend their kit lightly. In this one case, they seem to have accidentally made (for my use case, running 5 APs at home) a perfect product. They're cheap, extremely reliable, my wife doesn't hate them (though mostly they're in the attic or basement; only one is visible), they've got a (relatively) easy to use UI that manages all of them at once, and (Except for the switch 2) they seem to just work even though I've got vlans and lots of SSIDs and other goofy stuff).

If I had a simpler house to support, I'd just get a single WRT capable "big fast" router / AP...

Unfortunately you need to use the cisco software / firmware. The access points run linux but they're locked down like crazy with signed firmware blobs and such.

That said, the cisco firmware for this specific generation of access points is actually free and trivial to get -- create yourself a cisco account and go to downloads and download the 3802 "mobility express" firmware. The last ME firmware came out in 2024 and all this equipment and software is now totally unsupported by cisco so don't run PCI transactions at home... I'd also avoid running their captive portal or some of their other weird features...

Actually setting it up is a bit of a chore but it is a full featured "enterprise" (cough) AP management system with all the knobs and twiddles you could ask for.

It's really only a good idea if you don't value your time (like me) or if you have a sprawling plaster house where you want to have lots of cheap access points instead of a couple super fast ones.

Lastly, for better or worse, I haven't been able to make my kid's switch 2 work on the network.

It is probably a combination of hitting a (low) cost and mimo; cisco makes pretty reasonable looking APs with lots of radios and decent coverage and they look like UFOs not alien spiders.

But it's probably easier / cheaper to get maximum coverage at larger distances from a single AP using a big array of sticking out antennae, and that's what a normal home user is going to want.

I have been extremely happy with cisco 3802 access points purchased on ebay for $25 each. Sure, it's only wifi5, but they're pretty solid and you can just deploy a swarm of them.

And they don't look fugly.

It is a tremendous shame that cisco hasn't opensourced / unlocked this generation of kit.

Oracle makes their money by promising the C Suite "give us money and we'll solve your problems." There aren't actually many companies in a position to credibly make that promise.

The oracle database is also actually pretty nice, according to the people I've talked to who use it.

But mostly they play in the 'nobody ever got fired for suggesting' club. These days AWS is the most egregious landlord of that club, but whatever.

I think the "party line" is that the goal shouldn't be "1950s grid, but bigger and with more red meat!"

Renewables actually cover a lot of the newer demand and can be put around where it doesn't require as much transmission capacity because the production is happening closer to the demand.

And frankly, these jumbo sized data centers can just go figure out how to power themselves. If they want to put PV and wind and some peaker plants on their campus and they can actually meet the environmental requirements, go to it. So far they seem to be dumping AI dioxin into the local waterways, but it is probably possible to run these things without grid power.

In my whole career I've seen computers getting smaller and more efficient. Sometimes you'd need more of them, especially if your product is growing, but almost never would you need these radical increases in power draw to keep up with the red queen.