Anyone with experience plotting to tape? Someone has to of tried this. Do the excessive read times matter? Certainly it could be optimized to fit within 2 minutes to playback the correct data.
Say. 9,000tb on one simple tape……
Tapes Been hiding out quietly not getting any more advanced(smaller data)
All the while hard drives have about hit their cap on what’s physically possible out of a spinning disk. (Can’t get smaller data)
Tape isn’t bound by the same constraints as magnetic spinning disks.
If the same technology is applied to tape. Chia jackpot.
Yes, people have tried. The delay that it takes to pull a file from tape is way way too slow. During the initial design of Chia, they purposefully made it so tape would not be an option.
I’m not sure what this means? Are you referring to the maximum capacity of the HDD hitting some barrier? There are sample batches of 32TB drives at some datacenters right now… so I’m not sure what you mean with this and how that means tape is more viable than HDD? (they plan full scale production later this year)
The reason tape isn’t viable is the bits of data you need could be on the opposite end of the tape (depending on model thats quite a few minutes, the longer the tape the longer the delay) and thats assuming you could switch cartridges rapidly enough, some of them take several seconds to unload the last and load the next before you even seek data. Meanwhile HDDs can rapidly get the scattered bits sometimes on a single rotation of the platter.
I think we all thought of these fun things, I’m sure you’ve probably thought about, what if I take one of those 500-disc CD carousels and replace it with a DVD unit so I could get a bunch of plots taking near no energy to store, and what if I could ramp that up to some industrial quick retrieval mechanism… we’ve all been there, its part of the fun ways to think about ways to try and do it.
P.S. Ultimately we’re headed towards nand (SSD) being the beast, its getting more dense and cheaper, that will ultimately win out with its tremendously lower power between signage points, a few watts of idle power to host petabytes of space, Chia just gets greener as technology advances.
Super para magnetic limit has Ben reached for spinning disks… sure they keep starching like making smr drives. But the limits are what the limits are. They arnt getting much better.
Tape. Is the future of big data storage. Why not chia storage. For mega cheap in comparison.
Sure they have right now 100tb ssds. That fit in a 3.5 inch for factor. But we’re talking about 9000tb in a 3.5” form. There’s no comparison. Tape sales have Ben climbing its only a mater of time.
Sure this isn’t consumer focused. But say u raid 10 100 tapes with 9,000tb each. Read speeds would be fine I’ll bet. Sense that’s all that matters here. Then u win at chia. Period. Full stop. For a mear fraction of the price. CDs have thier limits just like spinning disks.
I watched that video (yesterday at lunch). But he clearly says that he sees it to be used for stuff like google’s archive data and stuff like that (in the future). Not things that need immediate access. So, basically bringing tape into the main storage level instead of backup level. But it is still tape. It is still slow. Great tech. I like what they are doing. But until there is another leap in the technology (which just may happen sooner than later), it isn’t a replacement for the majority of hard drives.
I think the next step in spinning disk drives is larger physical drives. Just look at what has been the standard physical footprint for so long. They could easily make them bigger (wider, thicker) and gain a lot of space. If the limit has been reached and they still want to make bigger drives, that will have to become an option sooner or later.
Sometimes I wonder if people read what was replied and took a moment to absorb it before repeating the same thing they already said.
I’ll try again, slower, like a tape…
it can take MINUTES just to reach the file you want on tape, little alone go back and forth once you reach that section of data to retrieve the random parts of it, this type of latency is bad for Chia because you have a limited amount of time to respond to the network.
Now JUST to entertain you, you could use really small tapes and a LOT of heads as a result, but at that point, you now have a system more expensive in initial cost and ongoing costs compared to a hard drive, heck, these days a system like that would cost more than an SSD… so you lost the entire benefit and cost savings of tape trying to make tape something its not, thats why its infeasible.
(hopefully that helps explain it, again)
Your suggestion “why don’t you just RAID10 100 tapes” tells me you don’t understand how tapes work, or how RAID works, or why using massively sized tapes in massive quantities results in what I just said above (not only would you still have the massive latency issue, but now you have extra processing to conduct on top of it, you literally made it worse than it was before).
If you want to find an example of massive tape storage and how long it takes to retrieve simple files much smaller than plots, you can look at Amazons fleet of cold storage, if they haven’t figured out a solution to the latency at their scale and need, why do you think its possible for you, they would stand to benefit billions from such a solution, if you come up with the solution, Chia will be your last customer, not your first, because you would make billions selling it to Amazon and the rest of the world first.
ur a troll **** off my thread troll, you don’t like it don’t read it, bye. no one asked you a thing.
here everyone saying YET again how unhinged Dr Jones is for his crazy ideas, being a trol,
you’d think you would of all learned by now.
i called out plot compression months before it was a things… was told how crazy and absurd it was then… till it was done…
i criticized this forum 2 years ago for the incoming spew of complete bs…
i warned everyone.
again was chastised.
and now
my moneys on tape.
i promise this will be optimized.
mark my words first. again…
peace out
in 10 years, when chia still hasnt blown up yet… and 100 tb seems like just absolute nothing… compared to “big data” think of the rate all our data is growing at… it is still exponential…
if ibm JUST LAUNCHED a new machine that runs on tape… eventually this will leak down to consumer technology. quickly i think.
what if the tape was made extra fat. instead of long.
and what if instead of “tape” its thin carbon fiber sheets. that spin at 15k rpm reel to reel.
give it 20 years… when were all literally drowning in data… and our tiny little chia farms are just dust in the wind.
you really think they’re gonna do away with the 3.5’ hdd standard? just eventually have to right? careful with silly ideas like that. could get you called a troll…
if big data is gonna grow exponentially thanks to this new tech so should chia.
what are we all gonna do when they move onto k34 only, or whatever. were gonna need vastly more storage. js
playing with chia… this is the long mans game… im thinking in terms of 50 years with my farm. everything evolving.
i have a friend working on cracking the tape code right now peer say… how to optimize exactly, in software. a new filesystem. and at hardware lvl.
is why im such a firm believer… thought id share the message. cause its in the pipes for sure.
just because im light years ahed of most here doesn’t make me a troll.
soo thanks for the response.
Just for anyone that believes the nonsense posted in this thread, hard drives are far from hitting the cap on capacity. Anybody can buy a 22TB drive now, 30TB are being evaluated now, much bigger drives are on their way.
If you look at the road map 50TB drives are planned.
Someone once told me back in the nineties broadband had reached the maximum speed possible, history will tell you the rest.
