Spinning Rust Ain't Dead Yet [CL72]

Posted on Tuesday, Sep 5, 2023 | Series: Chaos Lever
The humble hard drive isn’t going anywhere, despite rumors of its imminent demise. Tape too.


[00:00:00.090] Ned: Feels to be something on guy who’s late a lot. Oh, that was a deep cut of a sunny day real estate album reference, which you probably wouldn’t get, so doesn’t matter.

[00:00:14.690] Chris: If you don’t mind, I’m just going to smile and nod and back away.

[00:00:18.080] Ned: From you can’t claim to be an indie music snob and just completely miss oblique reference differences.

[00:00:26.810] Chris: I never claimed to be an indie music snob. I’m just a snob general.

[00:00:31.450] Ned: You’re not specialized in any particular way?

[00:00:34.730] Chris: No.

[00:00:36.170] Ned: I don’t know if that’s true. I feel like you’ve specialized in a few things to be particularly snobby about.

[00:00:43.390] Chris: I am very special.

[00:00:46.030] Ned: Everyone says so. Even your cat could not have timed that better. Honestly, it’s not going to stop either.

[00:00:56.530] Chris: Is the fun part.

[00:00:59.890] Ned: And what is your cat in particular looking for? Because I know when my dog does that, she wants, like one of two things and they’re both food.

[00:01:09.110] Chris: When your dog meows?

[00:01:11.670] Ned: Yes.

[00:01:12.330] Chris: Are you sure you have a dog? Because that’s a dead giveaway. You might need to review just categorization.

[00:01:26.730] Ned: Do you recall the Nicktoon cat dog?

[00:01:31.950] Chris: I recall it by name, yes.

[00:01:34.370] Ned: Did you ever actually watch it?

[00:01:37.870] Chris: It’s possible, but I don’t think so.

[00:01:41.470] Ned: There was definitely a period of time. I don’t know if it’s still true of Nick, toons, because I’m a grown up now. He says no one else is going to say otherwise, that it seemed like everybody at Nicktoons was on some kind of psychedelic. This is the era know, Rocco’s modern life. You had invader. Zim was a thing. And you also had Cat dog, just literally just half cat, half dog. Both the heads connected in the center, which really begged the question of what happens when they.

[00:02:22.670] Chris: You know, after.

[00:02:26.830] Ned: Yes, that’s kind of what I was getting at. It’s not quite a human centipede type scenario, but still this is a family show. Is it? What kind of family? Manson family. I don’t know. All right. Hello, alleged human, and welcome to the Chaos Lover podcast. My name is Ned and I’m definitely not a robot. I am a real human person and my central processing unit is not made of metal. Just like all humans, I’ve evolved past that. My brain is perfectly capable of 100% recall of all of my experiences, positive and negative. Especially negative, which is great. I think it’s something to do with electrons. Very, very sad electrons. Really the only kind of electrons. They’re all so negative. It but that was the joke, right? Yes.

[00:03:24.120] Chris: I appreciate you hammering the visitors over the head with it.

[00:03:28.100] Ned: Visitors? Do we have visitors now?

[00:03:30.850] Chris: I tried to say listeners and viewers at the same time and I ended up with visitors. It’s early, leave me alone.

[00:03:36.690] Ned: At least it’s a word. And you didn’t combine listeners and viewers into viewers or something.

[00:03:45.270] Chris: Lisseverts.

[00:03:47.930] Ned: That sounds like either a disease or something that you get arrested for.

[00:03:52.970] Chris: They need a ZPAC immediately.

[00:03:54.580] Ned: In either case, somebody needs a ZPAC. Maybe it’s me, I don’t know. I am about to be infested with all kinds of germs and viruses since today was oh, that’s right, because school’s back. School’s back in session. Which means my little minions have gone off to get infested with things which they will then bring home and smear it all over the house. Love it.

[00:04:22.550] Chris: Something to look forward to.

[00:04:24.100] Ned: Yes. So you can expect previously we had.

[00:04:26.310] Chris: Salmonella, this year we’ll have Salmonella 3D.

[00:04:32.070] Ned: Now with smell of vision. Let’s talk about some tech garbage, shall we?

[00:04:42.970] Chris: Anything better than this, really, he says before looking at the topic and remembering we’re going to talk about hard drives.

[00:04:49.140] Ned: Woo. I’m strangely excited about this. I really want to see where you’re going with the whole thing.

[00:05:01.010] Chris: Before we go into it in depth. I will just say that this could have been many hours longer than the.

[00:05:09.110] Ned: Six we have planned.

[00:05:12.630] Chris: I hope you’re all sitting comfortably. Let’s begin. This month saw Samsung announcing some frankly absurd upcoming SSDs among them, and leading the charge in absurdification is a 256 terabyte SSD. Let that sink in for a minute. And a PB SSD solution that encompasses, you guessed it, petabyte sized SSD solutions.

[00:05:50.210] Ned: Okay.

[00:05:50.930] Chris: Now, we will get into the nuance of PB SSD in a minute, but damn.

[00:06:00.290] Ned: Damn. It almost begs the question when they say an SSD, how are they defining an SSD as opposed to a bank of separate drives that you combine together into an SSD? And I guess that’s probably what you’re going to get into.

[00:06:15.270] Chris: That’s part of it, yeah. So just to be crystal clear, the 256 SSD is one drive and the PB SSD is more of a disaggregated solution of many drives mushed together.

[00:06:30.810] Ned: I guess if you look close enough to a single drive, it’s still a bunch of chips that are functioning together as a single drive, correct?

[00:06:40.220] Chris: Yeah, that’s literally how SSDs work.

[00:06:42.420] Ned: Yes.

[00:06:42.690] Chris: And we will get into that a little bit as well.

[00:06:44.660] Ned: Okay.

[00:06:45.320] Chris: Now, if you’re done with the spoilers.

[00:06:47.690] Ned: I don’t know, continue, I suppose. Go ahead.

[00:06:52.450] Chris: So both of these are very future looking. Even in the Enterprise, we don’t have anything approaching a 256 terabyte single disk at the moment. Obviously these are both Enterprise products because one thing that was not addressed at the event is why a customer would want or need that kind of storage at all, ever. The market at the moment barely has any available options for customers at four terabytes, let alone larger. Still in the Enterprise, this is big news and announcements like it have led companies such as Pure Storage to announce the effective end of the existence of HDDs, aka old style hard drives, aka Spinning Rust, as soon as 2028.

[00:07:44.950] Ned: That’s cute.

[00:07:49.030] Chris: Yeah. It goes without saying that Pure might have a vested interest in saying such things since they are an all flash storage array vendor. Yes, still, they were pretty dramatic about it and have been for some months back in May. Sean Rosemary VIP. Yeah, he’s probably super important, but his title is VP of R and D at Pure flat out stated, quote, no more hard drives, meaning HDDs will be sold after 2028. Wow.

[00:08:25.890] Ned: Yeah, and we’ll never need more than 1 memory.

[00:08:31.890] Chris: 64 KB ought to be good enough for anybody. Or is it 640?

[00:08:36.020] Ned: I think it was 640 KB, because the other 300 whatever, was reserved for the operating system somehow. Anyway, it doesn’t matter.

[00:08:45.690] Chris: God, we’re old. So he didn’t just say this and flit away. He actually did give some reasons. The first one was power consumption, and again, he was high on hyperbole his own supply. Replacing HDDs with SSDs could, quote, reduce the power consumption by 80 or 90%. Okay, all right, this is one of the areas that I didn’t go into it, but suffice it to say, people have looked this up, and no, that’s not accurate. SSDs are better because there are no physical parts. Nothing has to move, sure, but it is not 80 or 90%. Sorry. The other thing he talked about was density. SSDs, he argued, were going to continue to increase their storage capacity at a geometric rate, while HDDs, constrained by physics, were going to fall behind. No questions were asked regarding why SSDs somehow existed outside of physics, but one has to assume Rosemary would have replied to any question along those lines with something like, jeez, bro, you know what I mean? Which we don’t.

[00:10:05.850] Ned: No, I was going to say I don’t know what you mean, bro.

[00:10:11.810] Chris: It’s always so fun to just you completely destroy somebody’s argument when you call them bro.

[00:10:19.590] Ned: Accurate.

[00:10:21.990] Chris: So he’s leading up to an interesting point. For both hard drive types, what is the upper limit in terms of their various form factors? How dense can these things actually get? Because they’re not infinite SSDs or HDDs, no matter what Pure would like you to believe. In fact, this is the reason that the Pbssd factor exists, because as you alluded to atop, it is not a hard drive in and of itself. It is a mechanism combining many SSDs into one coherent whole. Or as described when it was announced, I want to say 18 months ago, quote, an NVMe subsystem for disaggregated storage with greater rack efficiency for PB scale storage, unquote. Boy, do they have a way with words. But I mean, seriously, does this sound familiar at all?

[00:11:19.150] Ned: Sounds like a storage array.

[00:11:20.340] Chris: Brand new concept. Yeah, exactly.

[00:11:22.840] Ned: Feel like I heard of this before. Yeah.

[00:11:26.010] Chris: Now, interestingly, this is actually already how SSDs work inside the case. They’re just a bunch of little self contained hard disklets, which is a phrase that I made up. Copyright pending yay. All masquerading as one individual disk through the power of super duper fast, internal communications, and firmware. So it’s not a wild idea to say we’ve got this communication internal to the drive. Let’s expand it outward and enable you to create guess. Basically just get a rack plug X amount of hard drives into it. They all are fluent in Pbssd, and now all of a sudden you have petabyte sized, quote unquote, single drives, which when you talk about some of the use cases that they were talking about, is obviously the future. And of course, and I weep to say it, the example use case was AI.

[00:12:32.970] Ned: Of course it was, because what else?

[00:12:36.770] Chris: What else is there in the world at all? I don’t even think we have lawnmowers anymore.

[00:12:42.690] Ned: Well, I mean, they’re all AI powered lawnmowers, naturally.

[00:12:46.870] Chris: Oh, God. I’ve seen The Lawnmower Man. That does not end well.

[00:12:51.060] Ned: Oh, wow. I was going to reference the happening, but Lawnmower Man is way better as a movie and a concept.

[00:13:00.490] Chris: And this is another aspect that I didn’t go into all that much, is we’re reaching the upper limit of how much stuff we can mash into a 2.5 inch drive case or even a 3.5. But my side question, and the one that I didn’t go into in depth, is why in God’s name are we still using 2.5 inch drives in the data center?

[00:13:21.520] Ned: Well, importantly, we’re not. And maybe if you’d done your research, jesus, Chris, you would know that there’s a new form factor that’s becoming more commonly used, which is called the EDSFF, which stands for Enterprise Data Center standard Form Factor. Yes, I looked it up. It used to be called the intel ruler form factor, but then they opened it up as an industry.

[00:13:47.120] Chris: Remember the ruler? Those are so cute.

[00:13:50.290] Ned: They’re extremely adorable. And they basically run the length of the server or the depth of the server, and they fit into a One U size space and they’re hot pluggable, which is a step up from an internal style drive. And so within that, you can pack a ton of SSD chips inside that single ruler form factor. And it was the first time that I saw I want to say they packed like a petabyte of storage. And this was many years ago, eight years ago maybe, when they packed a petabyte of storage in a One U server using this form factor, they can pack a lot more in now. And it’s all using the same bus on the back, which is the Mvme bus, right? Yeah.

[00:14:38.170] Chris: If I remember correctly, ruler came out like seven, eight years ago.

[00:14:43.900] Ned: Yes.

[00:14:44.560] Chris: And the whole idea was to revolutionize the amount of storage that was possible in a pizza box server.

[00:14:51.150] Ned: Yeah.

[00:14:53.690] Chris: So, okay, cool. I did not see that when I looked it up, so I’m glad that it’s still around.

[00:14:59.630] Ned: Yeah. I’m trying to remember how I stumbled across this relatively recently. I think I was looking into adding some SSDs for one of my systems, and it’s not a server form factor. And I kept seeing this EDSFF two and E one. I was like, what is this? Will it sit in my computer? And the answer was definitely not. And you can’t afford it if you have to ask. Exactly.

[00:15:29.770] Chris: Okay, so in order to understand these drives and these crazy densities that we’re getting, it probably makes sense to talk about how SSDs actually work. Okay, so let’s do that. And again, I’m going to gloss over some things. SSDs are built and made up of a lot of what we’re just going to call cells. How does the cell work? It holds an electrical charge. I’m not getting any more deep into that.

[00:16:05.740] Ned: That’s fair.

[00:16:06.520] Chris: Just stick with that model.

[00:16:07.930] Ned: Okay.

[00:16:09.670] Chris: You put in as many cells as you can. That’s how much storage you got. Okay. However, SSDs have an ability to do something interesting, which is not just hold a charge, but read that charge. And this is the way that they get multiple pieces of information or a larger blob of information into a cell. These are listed out as SLC, MLC, TLC and QLC. And then there’s a Plc, but that hasn’t been released on the market. What they equate to is single level sell, multi level cell, triple level sell, quad level cell, and then finally penta level sell. Now obviously, when they released MLC, they did not expect any more density. We didn’t go single, double, triple, we went single, multi, triple. So that was a big Oopsie. But the problem was they printed all the labels already. You understand?

[00:17:15.200] Ned: We have all made naming decisions that we’ve come to regret.

[00:17:22.110] Chris: Yeah. Anyway, this really is the key difference between an SSD and an HDD. And this is where the pure guy was alluding to the fact that SSDs can have more data on them for a spinning disk. I think everybody already has an understanding. There is X amount of literal physical space on a spinning platter and there can be multiple platters per hard drive. They are physically written to by a stylus at the end of a moving arm.

[00:17:58.170] Ned: Right.

[00:17:58.640] Chris: It’s the record player model. The drive spins real fast. The arm reads the data. Either that changes it, deletes it, whatever. And the faster the spinning disk can spin is, the faster the drive can operate. Interestingly. And again, in things that I’m not going to go into more depth about, it is really not as much of a difference as you would think anymore. Because the biggest thing that makes drives HDDs run faster is in fact the firmware and the way that the data is logically deployed on the spinning disks, which is one of the reasons you don’t see 15,000 rpm drives anymore.

[00:18:40.790] Ned: 7200 rpm should be good enough for anybody. Damn it.

[00:18:47.690] Chris: This joke’s never going to get old.

[00:18:49.400] Ned: No, never. Just like us.

[00:18:54.650] Chris: Everything hurts. So back to SSDs, which have no moving parts, have to do it differently. They have all of those cells that hold an electric charge now, that electric charge has to be within an acceptable range. Too low and you can’t measure it accurately. Too high and you run risks of A, the electric charge jumping from cell to cell, which is bad, or B, nuking the drive, which is arguably worse. But still, this is pretty creative. When you started with an SLD, remember single? That’s the mistake that I made. I keep calling it SLD, not SLC. When you started with the originals, an SLC, a single level cell, what you were doing was saying, yes or no, is there a voltage in this cell? So you had a binary value of either zero or one. And like I said, the electrical charge has to be within an acceptable range and that range varies by manufacturer. But for this example, let’s just say 2 volts to 3 volts. Okay? The cell holds the charge, the cell reads the resistance and determines yes, no, is there a charge? That’s how an SLD works. This makes SLDs simpler to manufacture.

[00:20:31.060] Chris: It makes reading and writing extremely fast and it makes the cell much more stable. This is the big three things that make SLDs the best and unfortunately, also the significantly most expensive kinds of SSDs. It also runs into the same problem as the spinning disk we talked about earlier, which is you can only cram so many cells into an SSD. So what they did to change the amount of density of the drive is change the way the cell works. Instead of it being is there a charge or is there not a charge, they started to read the charge, which means we had an acceptable range of well, you can go from 0 volt to 3 volts.

[00:21:19.520] Ned: Right.

[00:21:19.710] Chris: Because obviously off is an option. So with MLD, which is, remember dual instead of single, it is doubled. So SLC can hold a value of zero or one. MLC can hold a value from zero zero to eleven. So if you’re counting in binary, it’s, it does this by changing the value of the voltage. It’s no longer yes, no, it’s a measurement. So you could say, and I’m going to read them all out here, this is a lot of numbers that’s going to make everybody’s eyes cross. Zero to 1 volt could mean a value of one to 1.65 volts could be a value of zero. One. 1.66 to 2.33 could be a value of 10, and 2.3 to three could be a value of eleven. Do you follow?

[00:22:20.480] Ned: I do follow. And I see where you’re going with this.

[00:22:25.070] Chris: As we go from MLC to TLC to QLC, the same thing happens. The only difference is the range of voltage gets smaller because again, you can’t go outside of the acceptable range or the drive will explode. I’m only a little bit exaggerating.

[00:22:45.450] Ned: Fair enough.

[00:22:47.830] Chris: So by the time we get to Qld, which is the most popular consumer level of drive, and the densest by far, we can now hold a value from in a cell that is the same size as one that could simply hold zero or one. It’s kind of cool, right? That’s a big change in disk density.

[00:23:13.830] Ned: Yeah, that is a big change. I’m sure it comes with some trade offs, though, which is why SLC probably still exists in the enterprise somehow.

[00:23:24.290] Chris: You are correct. QLC is way more data dense. It is also slower to write because you’re writing more to one cell and reading more from one cell and more prone to failure. Two reasons for this. First of all, as those changes in voltage get closer together, that means they have to be more precisely measured. And it also means that by changing the voltage in that way, it wears out the cells faster. Think of the margin for error when your options are either on or off versus a very narrow slice of voltage. Even the argument is the SLC wears out at the same speed as the QLC. But because the range of acceptable value is so much wider, it lasts longer at a high performance range. QLC, the second that that cell starts to slip, it’s written off. And you can’t use that cell anymore because it is no longer reliable. And after a certain number of these cells fail, the drive will mark itself as drive failed. This is worse the denser you get because the cell simply has to do way more work to do reads and writes. And this is inversely proportional to the data density per cell.

[00:24:49.070] Ned: Interesting. Cool, right?

[00:24:53.230] Chris: So how do we measure this? Well, there’s a couple of different ways to track drive failure. The one that is predictive and used in the market is called a TBW or a terabyte write. So this is effectively the expiration date for an SSD. How many times or how many terabytes of data have been written to the drive is an estimate of how long the drive can reliably function. Now, this is another game where you get into the market and you see 16,000 different hard drive options from all the hard drive vendors, right? So these numbers can vary enormously. So I just grabbed two different products from two different companies or two products from two companies. So four products, two plus two plus one plus two. One plus two plus two plus two. So a crucial P, three M two SSD, which I’m sure you know what all those mean, is a QLC drive and it is rated for 800 TBWs, whereas a crucial T 700, which is a TLC, is rated for 2400 TBW. That is four times the reliability difference in price is about 30%. So it’s interesting decision you have to make on that.

[00:26:21.370] Chris: But those are two different product lines. So that’s as close as I could get for those are the kinds M two S are basically the kind of drive that operates your motherboard, right? It’s in your laptop right now.

[00:26:31.250] Ned: Indeed.

[00:26:32.670] Chris: If you want something more standardized, a 2.5 form factor from a few years ago, samsung has a series called the QLC version is rated for 360 TBW and the TLC version is rated for 600 TBW. So even there 60% difference in endurance and reliability. And if you want to go into something more enterprise at that scale for that type of drive, the Pro version, which is MLC, remember that’s the Duo is rated at 1200 TBW. So as you go up in density, you go down in reliability in a pretty what’s the word I’m looking for? It’s pretty mappable and pretty obvious no matter what product you’re using.

[00:27:31.890] Ned: It’s interesting that it’s measured in writes. So it seems like if you primarily perform reads on the drive, then it should in theory last a lot longer than a drive that has a really heavy write balance to it.

[00:27:49.110] Chris: Yes. Not only that, if you do side by side comparisons of these types of drives, the read speeds are not nearly as dramatically different from one another. The write speeds are where it falls off a cliff.

[00:28:02.980] Ned: I see. So if you needed an extremely performant drive for doing writes, if you were putting something in, say, like a database server, and you wanted something for that heavy write disk, you would go with something like an SLC or an MLC for that. But for the long term data storage, a QLC might be fine.

[00:28:23.810] Chris: Correct.

[00:28:24.400] Ned: Or maybe you use a hard drive, for example.

[00:28:28.850] Chris: Interestingly, and the last point I’ll put in here, as we talked about at the top, an SSD drive is actually a combination of a ton of little drivelets that all masquerade as one disk. One of the things that these manufacturers do, the QLC, let’s say it’s a two terabyte drive, the QLC disklets will make up something like 1.9 terabytes of that. The other part of the drive is actually used as a cache and is generally SLC. So it tries to mask some of those write speed limitations by having a super fast cache in front of it, which then the drive transfers that data into the actual disk itself in the backend when it doesn’t have to use the cache. If you’re doing simple file transfers, that means that you never see the write limitations of a QLC drive again. This is where it falls apart in terms of the enterprise, where if you’re running fast writes all the time, that cache will get filled, forcing you to write directly to the QLC and performance falls off a cliff.

[00:29:34.690] Ned: Right.

[00:29:36.250] Chris: Which, incidentally, if you go back in time a little bit, is exactly what we did with hybrid drives.

[00:29:43.530] Ned: Okay, I remember those.

[00:29:45.180] Chris: Or fast catch.

[00:29:46.820] Ned: Yeah.

[00:29:50.030] Chris: Anyway, so where does that leave us with HDDs? So HDDs sort of have a different set of problems. The first one is they are way more sensitive to environmental concerns. They have moving parts. They do you ever drop an HDD?

[00:30:12.210] Ned: Yeah.

[00:30:12.870] Chris: How many beats did your heart skip? Did you start writing your letter of resignation before it hit the ground or immediately after?

[00:30:21.910] Ned: Oddly enough, it was on the hard drive to begin with.

[00:30:27.030] Chris: That’s the kind of thing that is much less of a concern with an SSD. You ever leave your laptop in the car in the winter and you had to sit there and wait for it to warm up before the hard drive would spin? These are places where SSDs not only won, but just demolished the competition.

[00:30:48.510] Ned: Oh, yeah.

[00:30:50.190] Chris: And people of a certain age might remember that the first MP3 players had little tiny spinning disks in them.

[00:30:57.170] Ned: They certainly did. You could feel it when you were holding the first gen ipod. You could feel the hard drive spinning up. And I didn’t have a first gen ipod. I had a Rio something. I don’t remember the exact model. Yeah, but I had a Rio, and that had a 20 gig hard drive in it. Holy cow.

[00:31:21.230] Chris: Whoa.

[00:31:22.000] Ned: So much music on that thing.

[00:31:24.170] Chris: Whoa.

[00:31:24.820] Ned: But, yeah, I remember when was it the ipod mini came out? That was the first SSD. Or it might have been the ipod nano. I think it was the no, the.

[00:31:34.610] Chris: Nano was after okay.

[00:31:35.870] Ned: I think it was the mini was originally small form factor hard drive, and then they switched it to an SSD in the second gen. And that was like a game changer, because now you didn’t have to worry about what would happen if you dropped your ipod, which people did constantly.

[00:31:55.670] Chris: I’m looking at one right now. Anyway, believe it or not, HDDs, though, do still have their benefits, and they are still valid in the data center and at home. In both cases, HDDs remain cheaper gigabyte per gigabyte than their SSD cousins. And in some cases, and at high density, that could be three to four times less expensive. This is one of the reasons that I have two HDDs that I use as my backup drives. Yes, plural. Be responsible. One of those drives is like, ten years old at this point, which does make me a little nervous. But hey, that’s why you have the backup to the backup. And the reality is, a 256 terabyte drive is insanity. The fact is, if that drive fails, you have now a problem with 256 terabytes of data. So a massive drive is always going to be a risk. Hence why Raid was invented back in the day and why PBSD is reinventing the wheel today, spread the wealth and limit the blast radius when a drive fails, because it’s not if, it’s when. Yes, entropy is a thing. Get used to it.

[00:33:21.630] Ned: Or don’t. Entropy doesn’t care.

[00:33:27.490] Chris: One other fun thing that we can talk about in terms of mass storage, reliable data preservation now and into the.

[00:33:35.620] Ned: Future no, don’t say it.

[00:33:38.950] Chris: Magnetic tape, baby.

[00:33:40.570] Ned: God damn it.

[00:33:43.910] Chris: Magnetic tape. Even as we speak on the market today has more capacity than any single disk drive of any type and almost infinite reliability to boot. Not actually, like literally to boot, because if you try to boot off, you get the joke.

[00:33:58.830] Ned: I get things mostly.

[00:34:01.030] Chris: Capacity forecasts show SSDs, HDDs, and tape all continuing to expand the number of usable terabytes with tape consistently on top, consistently even beyond 2028. Ned, just this month, IBM announced a TS 1170 tape that handles 50 terabytes native at an I O rate of 400 megabits per second. Which is pretty fast. Yes, like, period. It gets even better if you look at the full ecosystem, because it’s not just buying a tape. You have to buy a tape library, you have to buy software, et cetera. But if you do, you’re looking at 877 terabytes native with the conservative three to one default compression, leading to 2.63 exabytes of reasonably accessible storage. Now, that’s on the market right now. It’s important to note, though, that even though this is an entirely new system, because it’s a new tape type, new drive, new software, all that stuff together, it is only an incremental improvement on current technology. It is an impressive leap because that 50 TB negative native more than doubles the storage available on previous tech. And everybody’s favorite LTO nine. Back in the lab, IBM has been working with Fujifilm, and they have been teasing revolutionary technology for magnetic tape that will be ten times denser at native capacity.

[00:35:38.090] Ned: Wow.

[00:35:38.880] Chris: Holy crap. Now, they’ve been saying that it’ll be out by 2025 for like three years now, so let’s be a little skeptical. But it’ll definitely be out. And so will something from LTO. The LTO Ultrium Roadmap has 576 terabyte native tape listed in just five more generations, which, if you’re keeping score at home, is approximately five years. Okay, so, yeah, that’s not going away either, whether you like it or not.

[00:36:12.870] Ned: No, I mean, there are some obvious limitations to tape. If you’re not writing and reading sequentially, it’s a lot slower because it doesn’t really have random access as an option. No, that’s the point for things like archival of data or archive retrieval of data, where you are doing just straight up reads off of a tape or just writing straight to tape. I mean, that’s pretty hard to beat in terms of capacity, right?

[00:36:48.070] Chris: And one other thing to note is that tape has the ability to physically do write once, read many, which is a pretty big deal in the whole don’t get ransomware and lose all your backups argument. The way it’s done on hard drives of any type is in software, so it probably works just fine. But if you really want that level of reliability, you need a physical medium. Also, fun fact, LTO has distinctive two toned cartridges, so you can tell the difference between a worm drive and a.

[00:37:23.410] Ned: Regular OOH who’s fancy you fancy.

[00:37:27.710] Chris: And also, tapes don’t need power, like ever. Unless they’re being or read, which kind of ruins the other talking point.

[00:37:37.410] Ned: Yeah, I mean, they do need power in the sense that you need to store them in a climate controlled area. So there is some power required, but your point is taken. They don’t need to be actively powered on, though I’d argue an SSD that also doesn’t need to be actively powered on to pull data. But I think the big win here is the fact that you have these enormous tape libraries with a robot that can automatically retrieve a cartridge and the only thing that needs power is the tape drive and the robot.

[00:38:12.010] Chris: Right. And it’s a fun fact that tape makes a huge impact in companies that use things like the cloud, because guess what, backs up a lot of the stuff that’s in the cloud.

[00:38:28.370] Ned: Yeah, it’s tape, isn’t it?

[00:38:31.570] Chris: So to be fair, they’ve never officially released what goes into like glacier or long term storage, but we have a heavy amount of innuendo that assumes that tape is the majority of it, but they will use whatever they think is the cheapest. And for some data that hasn’t been accessed for years, that could literally mean just writing it out to a Blu ray. That’s an interesting side tangent that I also didn’t go down and then just looking future, future. There are some things that will supplant these technologies. Eventually it will happen, right? Some of them we’ve talked about on this show or some other version of this show. These include things like DNA based data storage, quantum memory, and of course, 5D optical data storage, aka the Superman memory crystal. All these are extremely pie in the sky at the moment, but if the history of it tells us anything, it’s that one of these things will make a massive breakthrough and become mainstream far sooner than we think. But it’s still not going to be like next week. The reality is the hard disk landscape is the same as it ever was. SSDs for portable devices and first line storage, HDDs for second line or less performant storage and tape for archival.

[00:39:56.260] Chris: That is not changing anytime soon and certainly not by 2028.

[00:40:03.890] Ned: I’m sure that that VP you mentioned earlier has a bit of a myopic view of the storage world, especially given the company that he works for, and it is in his best interest to talk about how hard drives are dead. Somebody else who has a tendency to be occasionally wrong may have said that tape is dead about five years ago and they were very wrong and it was not me.

[00:40:29.610] Chris: We still have the tapes, Ned. We are ironic.

[00:40:37.050] Ned: Oh, I feel that deep in my heart somewhere. Yeah. So I think it’s interesting to see how the storage market is today and how the storage form factors that we have, none of them have really gone away. But it’s not like every storage form factor is forever. I submit to you the humble Zip drive ahead of its time, but certainly not used by anyone anymore, except for probably some random nuclear silo that started using it in the 1990s.

[00:41:10.230] Chris: That was the big improvement over the ENIAC.

[00:41:15.990] Ned: The eight inch floppy or twelve inch floppy that they were using. They’re like we upgraded the Zip drives.

[00:41:21.690] Chris: We’re doing it.

[00:41:24.090] Ned: Oh, hell. All right. Well, hey, thanks for listening or something. I guess you found it worthwhile enough if you made it all the way to the end. So congratulations to you, friend. You accomplished something today. Now attune yourself to the universal vibration of all things and become one with the cosmos. Or go buy a Slurpee. It’s the same thing. Yeah. You can find more about the show by visiting our LinkedIn page. Just search chaoslever or go to our website, chaoslever.com, where you’ll find show notes, blog posts, general Tom Foolery, and the sign up for our newsletter. We’ll be back next week to see what fresh hell is upon us. Tata for now.

[00:42:02.770] Chris: Did you ever do the thing where you just hung out behind the 711 because you were the coolest kid on Earth?

[00:42:08.170] Ned: We hung out in front of the 711 because we were rebels. Wow.

[00:42:12.670] Chris: Or you were scared of the dark?

[00:42:15.290] Ned: Shut up.

Show Notes

Spinning Rust Ain’t Dead Yet

Episode: 72 Published: 9/5/2023

The Spinning Disk Hard Drive Is Dead; Long Live The Spinning Disk Hard Drive

This month saw Samsung announcing some frankly absurd upcoming SSD products. Among them are a 256TB SSD and a PBSSD solution that encompasses, you guessed it, petabyte-sized SSD solutions. These are obviously future looking, and heavily based in the datacenter. Not addressed at the event is why a consumer would want or need that kind of storage, nor the fact that the market barely has any options for said consumers at 4TB, let alone larger. Still, in the enterprise this is big news, and announcements like it have lead companies such as Pure Storage to announce the effective end of HDDs, or, in parlance, spinning rust, as soon as 2028.

Intro and outro music by James Bellavance copyright 2022


Chris Hayner

Chris Hayner (He/Him)

Our story starts with a young Chris growing up in the agrarian community of Central New Jersey. Son of an eccentric sheep herder, Chris’ early life was that of toil and misery. When he wasn’t pressing cheese for his father’s failing upscale Fromage emporium, he languished on a meager diet of Dinty Moore and boiled socks. His teenage years introduced new wrinkles in an already beleaguered existence with the arrival of an Atari 2600. While at first it seemed a blessed distraction from milking ornery sheep, Chris fell victim to an obsession with achieving the perfect Pitfall game. Hours spent in the grips of Indiana Jones-esque adventure warped poor Chris’ mind and brought him to the maw of madness. It was at that moment he met our hero, Ned Bellavance, who shepherded him along a path of freedom out of his feverish, vine-filled hellscape. To this day Chris is haunted by visions of alligator jaws snapping shut, but with the help of Ned, he freed himself from the confines of Atari obsession to become a somewhat productive member of society. You can find Chris at coin operated laundromats, lecturing ironing boards for being itinerant. And as the cohost on the Chaos Lever podcast.

Ned Bellavance

Ned Bellavance (He/Him)

Ned is an industry veteran with piercing blue eyes, an indomitable spirit, and the thick hair of someone half his age. He is the founder and sole employee of the ludicrously successful Ned in the Cloud LLC, which has rocked the tech world with its meteoric rise in power and prestige. You can find Ned and his company at the most lavish and exclusive tech events, or at least in theory you could, since you wouldn’t actually be allowed into such hallowed circles. When Ned isn’t sailing on his 500 ft. yacht with Sir Richard Branson or volunteering at a local youth steeplechase charity, you can find him doing charity work of another kind, cohosting the Chaos Lever podcast with Chris Hayner. Really, he’s doing Chris a huge favor by even showing up. You should feel grateful Chris. Oaths of fealty, acts of contrition, and tokens of appreciation may be sent via carrier pigeon to his palatial estate on the Isle of Man.