But great advice and the route I've personally taken :)
In fact they have 14TBs for $199.99 right now ($14.28/TB)
https://www.bestbuy.com/site/wd-easystore-14tb-external-usb-...
Actually, aren't they already using that?
If you're in the United States, electricity is cheap enough that you can pick up much older SAS drives for really low $/TB cost and have it be worthwhile.
For example, I bought a used Supermicro CSE-836 [1], which is like a 3U server chassis with 16 hot-swappable drive bays and a backplane of some sort.
The backplanes vary, but mine came with the BPN-SAS2-836EL1. I paid $300 in total for the chassis itself, backplane, dual power supplies, heatsinks, etc, along with a Supermicro X9DRi-LN4F+ [2] and two Xeon E5 2660 V2s as a bundle from someone in the 'ServeTheHome' classifieds section [3]. From there, I picked up a load of HGST 3TB 7200rpm SAS2 drives on eBay for about $10 each from a recycling company. And then 192GB of DDR3 ECC memory from the same place for about $80.
I also grabbed a couple less-than-production-ready 3.84TB U.2 NVMe drives on eBay for a little over $100 each.
I think if I were to do it again, I'd have gotten slightly larger, newer drives. These are all totally fine, but I started seeing ~6TB drives for about 3x the cost per terabyte, which would pay itself off quickly with the energy reduction. The other reason is that I ended up going a little overboard; I have about 56x3TB drives right now, which is a lot more than 16, so I needed to get a couple of JBOD expansions to put them in, each of which were like $250 -- if I had gotten fewer, larger drives, I'd have had another $500 to work with & be saving on energy.
Another thing I'd have done differently is get fewer but larger sticks of memory. I have a really nice amount of RAM right now, but the energy consumption with 24x8GB isn't worth the upfront savings compared to getting 16 or 32GB DIMMs.
All the storage is in OpenZFS on Linux. The 56x3TB drives are configured as 7 RAIDZ2 vdevs, so 2 drives each are for redundancy, and 6 for actual usable storage. This leaves me with a bit over 100TB of usable space. And the 3.84TB U.2 drives are mirrored and act as a "special" device (lol, literally what they are called) [4] to automatically store small blocks and ZFS metadata.
I am sure I could have done a bunch better, but, so far, everything has been lightning fast and reliable.
I am using ZFSBootMenu [5] as my bootloader. It's cool since it is basically a tiny Linux distro that lives in your EFI and comes with a recent version of ZFS, so you can store your entire OS, including your actual kernel and such in ZFS, and you can enable all sorts of ZFS features that GRUB doesn't support, etc.
This is nice because, since the entire OS is living in ZFS, when I take snapshots, it is always of a bootable, working state, and ZFSBootMenu lets me roll-back to a selected snapshot from within the bootloader.
The Supermicro board has a slot for a SATA DOM [6], which is sort of like the form fact of an SD card. I picked up the smallest, cheapest one I could on eBay for like $15 and use that to store my bootloader. I did this so that my tiny 128GB SSDs that I use for my OS could be given to ZFS directly for simplicity instead of having to carve out a small boot partition, etc.
All in all, I'm probably out about $1750 for >100TB usable, redundant, fast storage, and a decent bit of power for virtualization and whatever else. It costs me like $50ish a month in electricity because of all the drives and DIMMs. But I was already paying 65 euros a month for a 4x8TB server from LeaseWeb to use as a seedbox, and ran out of space, so it's been worth it, even with my dumb decision to use 3TB drives.
[1]: <https://www.supermicro.com/en/products/chassis/3u/836/sc836b...>
[2]: <https://www.supermicro.com/products/archive/motherboard/x9dr...>
[3]: <https://forums.servethehome.com/index.php?forums/for-sale-fo...>
[4]: <https://www.youtube.com/watch?v=QI4SnKAP6cQ>
[5]: <https://zfsbootmenu.org>
Have you put a power meter on that setup? I've always wanted to do something like this but electricity here is AU$0.27/kWh, and my house is poorly oriented for solar panels unfortunately.
So, give or take, with a stupid build like mine, here, it's like .7 * 720 * $0.118 = ~$60.
Another thing to note is that mine stay spinning instead of idling for much lower energy consumption, cause I've always read that the stop-start cycle is what really tends to kill the drives.
I think it could be a lot more affordable/justifiable with:
- fewer, larger drives (6-8TB each?)
- fewer, larger DIMMs (16-32GB each?)
- perhaps slower, more energy-efficient drives (5400rpm?)
Also, I have seen some of the Atom C2750/C2758 boards for fairly cheap here.
If you don't need an absurd amount of RAM/PCIe/CPU, that could be a good option to save another bit of power.
> and my house is poorly oriented for solar panels unfortunately.
And yeah, same, unfortunately. Ironically, despite being in Florida, I barely get any sun; my entire neighborhood is filled with massive trees, lol.
I'd still like to look into solar at some point when I've got more time & money, though. It's been awful the past couple of hurricane seasons because of the trees -- something always seems to fall and cause a bunch of damage, so I'd like to cut down a few of the scary ones anyway (I am sorry trees!)
I'm not sure if this is something you're doing already, but don't forget that if you zfs send a snapshot to your storage pool, you can boot from that snapshot in a pinch.
We also (this past week) added a modifier to zbm.prefer - !!. If you specify zbm.prefer=zroot!! on the KCL, it'll only import the zroot pool. This might help your boot times by skipping importing your 56 drive storage pool.
We should have a new release out in the next few weeks, otherwise that feature is in the master branch if you build your own EFI executable.
And thanks for the software! I've been using it on all of my non-Apple devices for a while now. And my mom has been using it on their computer for a fair bit as well. It's been a total lifesaver on a couple of occasions and streamlined a lot.
I'd encourage you to write more of this technical content!
I don't think people, in general, are anywhere near as vocal and explicit about the things they appreciate compared to the things they actively hate or that frustrate them. And I think this applies even more so in careers and hobbies like tech where people don't seem to think of the things you make as consumable or creative in the same way they do with others (e.g., movies, videos, music, and other art); my personal experience has been that if someone is reaching out, it's—more often than not—to let you know about an issue or complaint.
So, thanks, it felt nice! And thanks for the reminder; I should let others know the same more often too.
Man, even enterprise servers, sold directly to data centers, max out these devices at 64 gigabytes. Many are offered with a pair of 32GB SLC old-school SSDs (for the extra durability).
Using 3.84TB for this is an enormous waste unless you host half the GitHub yourself and have specified "all files under this or that size go to the special devices".
Better make a torrent or book or comics mirror or something on this pair of huge SSDs. I use a pair of 32GB extra-durable (SLC) USB pen drives for special/metadata on a fairly decent ZFS dataset (~9TB) and they have something like 4MB of taken space...
We are working on hosting this collection, as well as saving other large collections. Please consider supporting us, donation details at http://pilimi.org (we'll set up a patreon-like system soon)
[Max size <50MB]
Hosting in the US, by someone in the US is foolish if your goal is breaking US copyright laws.
I have steadily become convinced that we should make a multi-billion year backup of all of humanity's knowledge. All of it.
Something that,
- is resilient (can survive a nuclear explosion)
- requires no power
- doesn't require software to read/reboot from
- (theoretically) lasts for at least 1 billion years
Copying from the Long Now Foundation's projects, I think we can achieve these goals by miniaturizing pages and etching them on some metallic surface (a titanium alloy), depositing a layer of some resilient transparent material onto this surface, and creating multiple copies.A few copies for Earth. 2 or 3 for the Moon. And a few sent out of the solar system on probes like Voyager.
Some simple math, if the pages could fit inside of a 10mm x 10mm, then for a plate that's about the size of an average coffee table at 2' x 4', we could fit 7,432 pages.
Assuming that we have 50 billion pages, we'd need about 6.7 million such plates to fit all of human knowledge, so far.
It sounds crazy, but assuming we could get net costs per plate down to $500, each copy would be about $35M. Or, ~0.14% of an Uber. Alternatively, 0.002% of the F-35 program.
That's doable!
-
6.7 million plates will probably weigh a lot. So off-world copies might need to use an alternative encoding scheme.
Another problem is likely to be organization of the plates/copies. The hardest part might be putting it all together in a way that can be trivially decoded by human descendants, even if they don't speak our language or share our subspecies.
(apologies for any typos, it's very late at my end)
The main issue would be loss of language. You’d end up with a gigantic Rosetta Stone of indecipherable gibberish.
Better would be to convert all literature into hieroglyphics, or graphic novels (or both), and then etch them microscopically.
The Long Now Foundation has extensively studied the language loss problem and arrived at a fairly elegant set of solutions.
https://longnow.org/ideas/02022/01/20/linguistic-data-in-the...
Rather I think you vastly underestimate what a billion years can do. The earth itself, and all that was on it, was formed a “few” billion years ago.
Our rivers alone have carved entire valleys into mountain ranges in much, much less time. I doubt a titanium alloy and some unspecified sort of super epoxy stand a chance.
And constant custody with regular restoration cannot be guaranteed for billions of years either.
That’s a massive problem for one plate (and its many copies) alone, more so for millions of unique plates…
The Long Now Foundation is only shooting for 10000 years, as far as I know.
On earth clay tablets have done a great job : we have tablets 6000 years old, so we know that works. That’s 60% of 10k already. Titanium seems expensive and might be melted down in time of need, like bronze has been often in the past. Clay tablets survived partly because it’s a ‘worthless’ material.
The ones that survived
Just don't make them look gold (even if they're not solid gold), or they'll get looted and melted.
Though it would be fun if a murderous warlord ended up with a gotten crown made of the Wikipedia article for toilet paper orientation. [1]
[1]: https://en.m.wikipedia.org/wiki/Toilet_paper_orientation
In the next few hundred years it's pretty likely that gold price will be marginal, as asteroid miners looking for less useless materials such as nickel, cobalt and platinum mine an excess.
So no, unless for some hidden reason asteroid mining doesn't work out, which is pretty unlikely given how badly the developed world needs that material.
For the Moon and Mars, carve them into stone there. You'll have to defend against meteor strikes so they'll be carved and stored deep underground. No idea about seismic activity on those bodies.
For outer space, maybe it's pointless because finding, boarding, unloading an interstellar probe and sending the cargo back home is not easy unless you have very advanced interstellar ships. An orbit around the sun could be an easier place to spot. The Library of Ceres or the Library of the Troians?
Orbiting the sun seems like a pretty stellar solution to the whole "keep something close and visible but safe without maintenance" problem.
Plus, think of how exciting it would be for a future civilization to discover that an object orbiting the sun was not a lump of rock!
But of course you need some marker then on the ground so people know where to look for it later if they want to read the data. Maybe some artwork? A giant monolith perhaps?
Perhaps at the D&M pyramid?
https://www.jpl.nasa.gov/images/pia04745-the-cydonia-dm-pyra...
https://en.wikipedia.org/wiki/5D_optical_data_storage
5d is a bit dramatic but it is a femtosecond laser writing of quartz crystal that should, in theory, be stable for billions of years and can hold hundreds of terabytes.
Seems to be just what is needed.
https://www.scientificamerican.com/article/data-saved-quartz...
The earth will be uninhabitable in that time frame due to changes in the atmosphere and beyond that the sun itself will complete it's lifecycle.
I know there's a big fad to "just believe" in a SF future that spans space and time but physical realities in this area are pretty rough.
I think the best thing people could do is realise that, eventually, everything ends and believing otherwise when it comes to the human race is much like believing in an afterlife.
Imagine if you were an alien species who somehow comes across this capsule hundreds of millions or billions of years from now. It's proof of sentient life elsewhere! But then you date the U-238 and realize that they're probably all dead...
But, they've left all of their civilization, culture, heritage, and knowledge behind. And you get to experience that, even recreate a tiny simulacrum of their world. And it gives you something, it's a tangible form of communication and cooperation across aeons.
-
In the short term, the backup is probably useful to have. Imagine if a collection is lost to fire or some other catastrophe, and one of the closer ones could be used to bring it back. Case in point, the fire at Notre-Dame. In 2015, Dr. Andrew Tallon, an art historian, painstakingly scanned all of Notre Dame, https://www.forbes.com/sites/trevornace/2019/04/16/we-have-b...
I think it's important to be honest. Would you have flagged his work as pointless in 2015? After all, the Cathedral has stood for centuries, been photographed so many times, what's the point of a 3D scan?
After 2019's fire, those scans became important (unsure to what degree) to the restoration effort. At some level, by capturing and preserving history, Andrew Tallon helped save history. And now it becomes a part of the story.
That's the goal. To engage in an act of optimism for the betterment of us and all of humanity.
We can barely get to the Moon and have no realistic prospect of making a self-sustaining Mars colony any time soon.
How is it credible to claim we can build starships that can cross lightyears without technological or sociopolitical problems?
The engines - which we're nowhere close to creating - are the easy part.
The real question is, is it worth making it a priority?
It will follow by itself once space mining kicks off. I'm sure Jeff Bezos would like to mine some asteroids.
That kind of claim needs an actual source, because it's not.
The distances are so many orders of magnitude too high for any of today's technologies, making this a pure daydream.
The best we could plausibly do is travel within Sol, and that's not starfaring.
It would have to be a multi generational journey in which any refueling/restocking is impossible and any slip up would cause a full wipeout.
Starfaring is far beyond us... and that's unlikely to ever change for humans with a biological body.
At that point someone can just put it on a thumb drive and take it with.
It's not beyond the realm of possibility that people ultimately inhabit a spaceship that wanders the universe looking for a new home.
We are doing that right now.
> The hardest part might be putting it all together in a way that can be trivially decoded by human descendants, even if they don't speak our language or share our subspecies.
Though other crystals should be highly stable. Quartz seems geologically and chemically stable, plain old glass ain't bad. Something reasonably cheap is probably preferable, both from the cost basis (an expensive-to-create archive is a challenge) and the repurposing challenge (a diamond-etched bibliographic archive might have other appeals to those who chance across it).
Even parchment proved sufficiently valuable that works were often repurposed (and lost) through palimpsests.
isn't the same idea viable if applied to silicon crystals (which we already grow lots of)?
And I'm over here hoping humanity extinguishes itself ASAP so the rest of life can thrive before we kill it all.
Intelligent life seems highly overrated.
TLS is not only for hiding the content, it's also for authentication: it ensures that no malicious middle party can modify the content, for instance to inject malicious Javascript (for an example of this happening, read about the "Great Cannon" attack on GitHub).
Additionally working on a UI for libgen that is far better as an open source project might be worthwhile.
> Z-library were in it for commercial gain (you could access a certain number of books but to get more you had to pay for a subscription). They started out as a fork of library genesis, whose mission has always been strictly non-commercial and about providing free access to everyone without limits.
What is the difference between what "z-library" was doing, which was "you download a lot for free, but if you need more, just pay us, because no internet bandwith is for free" from donations? I always understood the subscription at z-library as support, not different than patreon...
https://www.regulations.gov/comment/USTR-2022-0010-0024 https://www.regulations.gov/comment/USTR-2022-0010-0015
Ick.
I'm sympathetic to piracy, but the moment people start to make money off of pirated works it starts to feel much more wrong.
if youre worried about artists and authors, you got no idea how much cut publishing houses and music labels take from them.
So, an artist eeking out a living on something they worked hard on and released a month ago or two years ago gets sympathy from me. An artist who released a thing 20 years ago and still wants to eek out a living off that, such an artist doesn't really have my sympathy. The great-grandchildren who want to get paid for something an ancestor released 120 years ago[0]? Hell no, go do something yourself to make money.
If you're Lars Ulrich in 2000 and sue Napster over songs you released 10-20 years ago, while sitting on a net worth of maybe around 100-200M USD (now 350M), then my sympathy is with the pirates.
This doesn't touch the issue of how a lot of artists are not making much money, not because of pirates but because of predatory music labels and publishing houses.
It also ignores the problems with the "lost sales" theory. A lot of the pirates would have never paid for stuff they downloaded in the first place. And a lot of pirates started paying after pirating some stuff. E.g. I remember discovering a lot of artists from songs I illegally copied on LAN parties back in the day[1], usually artists too small to be on a lot of rotation on radio and MTV (yes, MTV used to have music, crazy) which I probably would have never known about otherwise. And I gave lots of money to these artists, buying their CDs, going to their concerts when possible, and so on. In the same vein, I discovered artists on whatcd and similar pirate places later on.
And doesn't touch on the "please take my money" issue... There are a lot of things that are out of print, etc, that you cannot pay for. E.g. large companies holding licenses to content may even take works out of print deliberately in tax avoiding schemes - I am not a Hollywood accountant, but from my limited understanding they can declare a loss when doing so which is worth a lot more in tax reduction than keeping a title in print.
And it also doesn't even touch humanity's need to preserve important cultural artifacts for the future.
[0] Remember, up to artists death + 70 years copyright term. While this exact scenario has probably not happened yet, as these rules are too new, you get a glimpse of what will happen in the future when you look about all the legal fighting still happening over Sherlock Holmes - a figure and body of worked created mostly before copyright law even existed.
[1] Yes, I am old. If you have no idea what a LAN party is, it's basically a bunch of people actually meeting in some venue with their computers, wire everything together into a temporary LAN, to play games and swap files, which back in the day really was the only sane way to do mulitplayer and filesharing stuff as internet speeds were so limited, and internet was usually very expensive, often still paying by the minute.
Creators being unable to earn income is a problem - but artificial scarcity is not a good solution, as it benefits the few at the expense of many. Imagine if everyone was prohibited from sharing news over internet because "free distribution of news harms the journalists who rely on newspaper sales" - that'd be absurd, wouldn't it?
Old-school copyright relied on the natural scarcity of paper and other distribution resources - digital age has lowered the distribution cost to zero, so the old model does not work anymore. Another model of rewarding authors is needed, one that does not rely on restriction of distribution.
BAM you just blew my mind at quarter to 8 on a Friday.
Scarcity of resources has nothing to do with scarcity of artistic works. Literature isn't fungible like a commodity, and can't be grown or mined as needed. It's also not a professional service or a form of labor. You can't put a gun to a farmer's head and make them write something brilliant, the way you can make them farm potatoes.
One could just as easily argue that society exists to organize labor in a way that increases specialization and efficiency. The reduction of scarcity is just a side effect. Specialization breeds scarcity in every new speciality until it becomes universally reproducible. Art is the forwardmost outcropping of specialization - it exists in advance of what it describes being known or understood - and by definition it is always the most scarce speciality. The artifice in making distribution of it remain difficult is therefore an extension of the natural place of scarce ideas in a world of abundant things.
If so, then what is the purpose of specialization and efficiency, if not for the reduction of scarcity, in terms of individual access to resources? If cooperation wasn't beneficial for the individual, nobody would cooperate (unless forced to do so, but I personally wouldn't want to live in such a society).
> The artifice in making distribution of it remain difficult is therefore an extension of the natural place of scarce ideas in a world of abundant things.
The whole idea of monetizing ideas the same way we monetize material things comes from their representation in scarce material things, such as paper. There's nothing "naturally scarce" about ideas - I'd argue the opposite, the ideas can be multiplied and shared at zero cost to the original author of the idea. The only cost comes from the way in which ideas are distributed. The golden age of music industry happened when the only way to distribute sound was either to perform a show (which was unscalable) or to buy a record (which was scalable) - producing a record was costly, but since there was no other way to listen to music at home, people paid. Companies charged more than production cost, and they made profit.
My whole argument is that treating manifestation of ideas as scarce things is an outdated view - in the digital age, all it takes to share an idea is a few button clicks. Trying to force scarcity will never work, and we need another way to reward the authors.
The price of a record was never just the cost of pressing the record. CDs cost pennies to burn; paper and digital printing by the late 20th century were extremely cheap. The wholesale and retail prices always included pay for the artist (along with a raft of agents and corporations along for the ride). Spotify itself imposes artificial costs to distributing music, in the form of limits, prohibiting downloads, subscriptions and advertising. They do this to make money for themselves, but also to pay the artists.
As you say, distribution is now essentially free from a technical standpoint. There's no philosophical difference between Spotify charging 100x what it costs them to stream a song, and a record company charging 20x what it cost to press a CD. Free distribution would mean zero money for artists. We started by talking about piracy. By definition, with piracy the artist gets nothing.
Since no one can sustain an artistic career and produce writing or music over a long period without some sort of income, this means either all artists will have to be from the wealthy classes, have other means of support, and simply want to make art as a hobby, or else there has to be a distribution channel with an imposed toll of some kind that ultimately funnels them payment for their work. Whether subscription-based, or charging per-device for copy protected media, that will never be a perfect system, will never be free of piracy, and will always appear to the end consumer as an arbitrary restriction on free information.
I don't think scarcity of paper has been the limiting factor on price since at least the 18th Century. Since there is no scarcity of electrons, I don't view scarcity as a useful paradigm. The prices of locked downloads are simply representing the actual cost and market value of the work, which should be the same whether on the paper or in the aether.
Artists should be rewarded for their work - but artificially limiting everyone's ability to share information is not the right way to do it. It would have severe consequences for the free society.
Another way must be found. I don't have any ideas, but I know that restriction of sharing information is not a good one.
Do not traditional libraries and their ebook lending systems, however inconvenient, provide the same service? All one needs to do is be patient until the book you are interested in becomes available. In the meantime, maybe you can check out a different book that is available that you might not have discovered otherwise!
[Edit]
Also check out OverDrive / Libby
"Piracy" is expensive! Most people do not donate to support these things it's understandable if they decide to charge a few cents or bucks to keep the light on.
Their policy was basically around discouraging datahoarders and scrapers as far as I could tell. If I recall correctly, the limit was something like 10 e-books per day per ip address. Not really a huge limit when you consider that there was no limit on file size for things like graphic novels. They never required accounts, and the only thing locked behind a required donation was compute intense tasks like conversion, and send-to-kindle.
The kind of bandwidth/storage that they had to be consuming is quite expensive and I don't blame the organizers for soliciting donations.
That's a silly policy in a world of shared VPN endpoints and Tor exit nodes.
The only thing approaching a dark pattern is asking you to sign in for advanced features. A free account gets you a bunch of those features.
The main page has a single 'donate' link in the upper right corner. A contribution as low $1usd was enough to unlock all features except higher daily download limits.
What dark patterns did you see?
- Various tricks to get the user to give an email (even if not necessary, similar to cookie boxes making "accept everything" very much easier than "only essentials".)
- After some time (download count or delay?) spam emails to try and get some payments out of the punter, again under false pretence.
Not saying it was bad value for money. Possibly not much worse than tricks average legal businesses employ. The point is someone sat down and devised that part of the UX with no other purpose than extract more money from users, by lying to them.
It never occurred to me that people wouldn’t realize that it was a piracy site, and not a “library”. I guess I’m just too tuned into the internet. Haha
I see your point now. It’s sort of disappointing that I'm so used to even worse dark patterns on legit sites that I don’t really see this as dark.
Practically speaking, if you are in the US, you could download stuff from zlib and get away with it - I believe that's not illegal. Hosting on the other hand is.
Using torrents to download is legally also hosting, albeit some random parts of what you are downloading yourself. So to stay legal in the US, you can't really use torrents.
Maybe some enterprising person in Russia or China can become a reseller of hard drives and preload them with chunks of libgen ?
I mean, it's sad that the greatest country on earth has a sneakernet gap with Cuba [1]! ;)
https://news.ycombinator.com/item?id=32972923
Z-library were in it for commercial gain (you could access a certain number of books but to get more you had to pay for a subscription). They started out as a fork of library genesis, whose mission has always been strictly non-commercial and about providing free access to everyone without limits.
Hopefully this will encourage more people to go back to the original libgen. I suspect Z-library's popularity was because of a better interface and larger collection, but I think lots of people didn't realize libgen offers all its books/papers for free without limits.