Grey Beards on Systems - 53: GreyBeards talk MAMR and future disk with Lenny Sharp, Sr. Dir. Product Management, WDC

Episode Date: October 27, 2017

This month we talk new disk technology with Lenny Sharp, Senior Director of Product Management, responsible for enterprise disk with Western Digital Corp. (WDC). WDC recently announced their future disk offerings will be based on a new disk recording technology, called MAMR or microwave assisted magnetic recording. Over the last decade or so the disk … Continue reading "53: GreyBeards talk MAMR and future disk with Lenny Sharp, Sr. Dir. Product Management, WDC"

Hey everybody, Ray Lucchese here with Howard Marks here. next episodes of Graybeards on Storage monthly podcast show where we get Graybeards Storage Assistant bloggers to talk with storage assistant vendors to discuss upcoming products, technologies, and trends affecting the data center today. This is our 53rd episode of Graybeards on Storage, which was recorded on October 23rd, 2017. We have with us here today Lenny Sharp, Senior Director Product Management responsible for Enterprise HDD and Product planning of WDC. So Lenny, why don't you tell us a little bit about yourself and your company? Sure, thank you Ray. Yes, I am with Western Digital. You got my title correct. I am responsible for the enterprise hard drive product line here at Western Digital and as you may know, Western Digital is the leading company in the storage industry right now with both a full product line on the HDD side as well as SSD.

And we've been experiencing quite a bit of growth. And I think the reason I'm here today is we've had a pretty exciting announcement relative to the future of hard drives. What is the future of hard drives, Lenny? Well, the way we see it is because data is expanding at such a rapid rate and companies are seeing real value in storing that data, there's an economic imperative to store the information the most cost-effective way possible. And for the foreseeable future, hard drives are that mechanism. So that's the vehicle that is working today. And I think based on the technology we just announced,

it's going to continue to be the best value proposition going far into the future. But people keep telling me that your cohorts at what used to be SanDisk are going to stack Flash four million layers deep, and it'll be power too cheap to meter. Why do we still need hard drives? Well, again, I think there's certainly a very, very important place for Flash and we're not disputing that at all. And my cohorts at Sandisk now we're under the same flag and working for the same team. the overall problems from the two different perspectives. But, you know, again, based on the announcement that we just made, you know, hard drives are going to continue to maintain about a 10x dollars per terabyte advantage over even the most deeply stacked, you know, QLC

NAND. And that kind of value is just going to be sort of irresistible for companies wanting to store vast amounts of data. Perfect. That's exactly what I wanted to hear. Yeah, I think the key there is if you can retain or maintain that price differential, there's a place for storage or just storage for as long as it can be maintained. It's when that price differential disappears that there's a challenge. Yeah, my guess is it's about 5 to 1. 5 to 1? Yeah, I don't know where it is, but at some point, as it degrades. Somewhere between 2 to 1 and 5 to 1.

Yeah, it starts to become less and less viable. Yeah, I mean, and just to jump in, we've seen that kind of tipping point in performance enterprise hard drives and client. But I think in the capacity end of the spectrum, with, what, 50 terabytes and a 2.5-inch form factor. But it doesn't cost any less than the 4-terabyte drive on a 5-terabyte basis. Yeah, yeah. Let's talk about the news from a technology perspective, Lenny. I was on the streaming event, I think it was the week before last, right? Where you talked about MAMR.

So what is MAMR and what does it mean to disk storage or enterprise HDD? Sure. Well, technically MAMR stands for Microwave Assisted Magnetic Recording. microwave-assisted magnetic recording. And the significance, though, is that it is the technology that we've chosen to extend the cost per terabyte takedown of hard drives for at least the next several generations. We can't say for sure, but right now it looks very promising to, again, continue to drive the cost down. So as we were talking about before, cost really reigns supreme in terms of storing vast amounts of data. So if we can maintain a cost takedown that keeps pace with what we're seeing from the NAND side, then hard drives will continue to be relevant. So that's really the key for MAMR is to continue to drive that cost takedown, you know, again, far into the future.

Okay, well, we've had conversations a couple of times with hard drive vendors about the roadmap for the hard drive. And we've heard about bit pattern media and we've heard about heat assisted magnetic recording but mammer is new to me um how does mammer relate to hammer and is it the same concept and how does this get us higher bit density? Sure. Okay, so the issue with growing aerial density, which is just basically how densely we can pack the bits onto the disk, really is related to this concept that as the grains on the media get smaller,

which is what is required to drive up aerial density, the head also has to scale accordingly. So as that head scales, the right field that the head can generate diminishes along with the size of the head. And at the the same time as the grains on the media get smaller we have to make them harder to write so that you don't inadvertently write adjacent tracks or adjacent bits if you will so you've got kind of two factors working against each other we've got higher coercivity media and weaker magnetic fields we can generate in the head at the same time exactly so we've been talking about this concept that we need some sort of energy assist to to sort of tip the scale so we've got these these harder to write bits and we've got to add some energy to the equation in order to uh you know change the state state of those bits from one orientation to the other.

So I think, you know, I think you had mentioned Hammer. Hammer has been the thing that I think the industry has been most focused on because it does actually show some very promising gains in terms of the ability to grow aerial density. You know, the issue with HAMR, though, is that, you know, HAMR is heat-assisted magnetic recording, and it uses a laser to, in fact, generate the heat, which lowers the coercivity or the, I guess, the ability for the magnetic grains to be magnetized. Okay, I would have used coercivity. Yeah, yeah, yeah. So, again, the heat.

But it's the magnetic stickiness. Yeah, yeah. Basically, the magnetic moment, if you will, to tip it over. Right. So the issue with heat, I mean, you guys have been around long enough to know that there's probably no area of the computing uh mechanism that heat is a good thing so you know motherboards heat removal is a major project um just just as a an aside uh here in santa fe new mexico we're at 7 000 feet elevation and i have a few one u servers and a few old viridant pcie ssds

and they don't work together because the fans won't move enough air to cool them at this altitude gosh because the lack of lack of oxygen lacks a lack of air well it's like you know mostly lack of nitrogen, actually. But it's just the air is less dense and therefore can carry less heat. And it was right at the tipping point anyway. Right. So again, we have lots of examples where heat is not good. The surface of the media being heated anywhere from 400 to 700 degrees C in order to achieve that writeability. So consequently, Hammer has...

The guys at Seagate kept promising me, but it's such a tiny spot. Well, it's not really very many jewels. It's still heat that has to be dealt with. And the reality is that that heat has created reliability and quality issues with the sort of veracity of the writing process. So those are issues that just have not been solved. Now, will they ever be solved? We continue to invest in Hammer, and it may be that we will see a breakthrough in the future in Hammer similar to what we've seen on the MAMR side. But in the meantime, with MAMR… The technological breakthrough that somebody is going to have to write an acronym to make come out to be SICKLE is yet to be invented.

Good one. Good one. Hammer and SICKLE. I get it. Anyway, I think that the issue is these reliability of concerns have been a challenge for the industry. They haven't been solved. I think Hammer is something that's been, is a technology that has been promised, but not delivered in a, you know, manufacturable, reliable way that actually drives costs down. And in the meantime, we were also in parallel investing in MAMR. And as it turns out, MAMR doesn't heat the media. There's no heat involved in the writing process. And instead of a laser heating the media, we're actually using a spin torque

oscillator, which is embedded in the write head that generates a field, a microwave field, that has the same kind of effect. It takes high-coercivity media and makes it easier to write for a moment in time. We can change the state of the bit, and then it becomes hard to write again, which gives you that thermal stability that you need to maintain the integrity of the data on the disk. So I always thought microwave would actually heat as well. It does not heat. At the media, at the point of writing, it's essentially the same temperature as a standard PMR process.

The microwave is used to make the bit easier to switch, but it doesn't actually provide any heat. But it does that by moving some of the electrons to an elevated state, and it's so in spite of the fact that we think of microwaves alongside ovens, there is actually no heating in the microwave frequencies that we're using. Right, but the microwave heating in a microwave oven is because that frequency of microwave causes water molecules to vibrate, and the vibration causes heat. Correct. I get you and so you have to add energy to cause the magnetic state the magnetic moment to be reduced but by adding just the right frequency of energy you can do that with it by adding a lot less energy than the sledgehammer

there's a freaking and a different kind of right, that doesn't have the downside from a reliability perspective. Well, I mean, from a quantum mechanical point of view, they're both EM, they're just different frequencies. Okay. Wow. So does this change the... So BPI, TPI, those sorts of things? I mean, does it affect both of those? Or does it just affects the BPI?

No, actually, we do get a benefit across both dimensions, BPI and TPI, with MAMR. Now, as it turns out, we've had a lot of other technologies that we were able to use to extend our TPI capability. So, you know, it's sort of like, okay, we've kind of gone as far as we can go. We need something extra. And, you know, MAMR extends our TPI capability, but it also adds the BPI component that we really haven't had with some of the other technologies. And just to clarify, TPI is tracks per inch,

and BPI is bits per inch within a track, right? Right. So track density, bit density. How would hammer and shingling interact? Actually, either hammer or mammer can be used with shingle tracks, so they're not mutually exclusive. And in fact, we will be using our mammer technology for both conventional recording as well as shingled magnetic recording. And in that case, you'll get a density boost by overlapping tracks. Correct.

The TPI will be denser with SMR. Yeah, but SMR really requires software smart enough to deal with it. Oh, yeah, yeah. I mean, it's a different use case kind of thing. There's operating system requirements, et cetera, and or smarter drives, one or the other, as far as I can figure. But SMR exists today in a standard PMR environment, right? I mean, there are drives available from you and other vendors that support SMR.

Yeah, in fact, just a week prior to our MAMR announcement, we launched our 14-terabyte SMR drive. Yeah, yeah, yeah. I thought I read something in some of the announcements about glass versus aluminum substrates. Correct. Being another advantage of MAMR over Hammer. Yeah, one of the issues with Hammer is that, again, because of the high heat involved in the recording process, you have to use a glass substrate. So that's the first change.

Now, glass has been used and is used in our industry for mobile drives, but has not been used for the capacity drives that we're talking about. So there is a tooling up and also an added cost per disk. And then in addition to that, we use cobalt platinum media for drives today. Hammer drives will require an iron platinum media surface, which is, again, a new, there's new invention, new development that has to go into that. And from our perspective, that adds technology risk. So you've got sort of a new recipe for the media.

Yeah, and you use a really tiny amount, but platinum still sounds expensive. Well, again, the platinum is common to both. So that's not a factor. And honestly, we're not looking at the iron platinum media development for Hammer as necessarily an added cost, unit cost. But the glass is so you know you know as we considered you know hammer drives either from us or from competitors you know as we analyze the the cost component and what could be gained in terms of capacity you know we we came to the conclusion

that between the glass media and the laser and the additional complexity in the head design um you know the the aerial density gain that you would get from a first generation hammer drive is probably pretty equivalent to the added cost so what we're really saying is you'd have a a generational change and a new capacity but basically cost that would scale with the capacity you know net yeah but as we've already established you have to maintain a decreasing cost yeah exactly and and i think i think that you know in our industry we find that if you aren't you know reducing cost each generation, you don't usually get a high adoption. So I think that the, you know, if the reliability concerns are solved,

we're still going to see pretty slow going for a hammer adoption because you've got to demonstrate a clear cost benefit if you want you know widespread adoption okay and this is because the hammer requires heating the surface so hot that the aluminum deforms or because the aluminum conducts heat so well that now you need a bigger laser um i i think just in terms of thermal stability could be a problem, actually. Yeah, I think there's multiple problems. So, honestly, there is no path, as far as I know, to produce a hammer drive using aluminum substrate. So, everything, all hammer development is on glass, and that comes with an added cost.

Right. Mm-hmm. And we didn't even talk about the laser that would have to be created and configured and added to the head assembly. Is that how it works? Exactly. So there's an actual laser and a wave guide to focus the beam. And that is a, you know, it's a component that certainly doesn't exist

in today's hard drives. It's extra cost for every single head and then i think you could also argue there's probably some yield issues associated with that which are also going to tend to drive cost up so i mean again well aren't those things also true of the oscillation over thruster the spin torque oscillator spin torque oscillation over thrusters from buckaroo bonsai i have to keep those straight no um you know the the spin torque oscillator. Spin torque oscillator. Spin torque oscillator. Right, oscillation overthrusters from Buckaroo Bonsai. I have to keep those straight. No, the spin torque oscillator is essentially embedded in the head when we fabricate it.

There is no significant added cost for that. It's just a matter of what materials we use and how we layer them and how we actually fabricate the head and there's um it's not the same kind of uh thing as a a laser which is again a discrete component with added cost i always thought optics required hand holding or something like that in order to get them properly aligned and stuff wasn't a real automated approach to that, but that may have changed. Yeah. Given the fact that I'm now buying 10 gig Ethernet SFPs from China for 15 bucks, I think that's changed.

Yeah, you're probably right. Okay. So how far does MAMR take us? Yeah. You know, assuming none of the ugly things that can pop up their head that you haven't thought of yet do. Yeah, I mean, we've looked at various ways to, again, introduce MAMR and then extend it through further innovation. We believe we've got line of sight to somewhere between four and

four and a half terabits per square inch, depending on the design of the drive that you'd be looking at, you'd be looking at something like maybe 80 or 85 terabytes. There's also obviously things that could be done in you know form factor and number of disks so um you know right now we yeah i've i've wondered for a while why we haven't gone to a taller three and a half inch form factor i mean in fact back to the original three and a half inch form factor which was like half an inch more Z-axis than the ones we use now. So, Lenny, are you saying 80 to 85 terabytes per platter? No, no, 80 to 85 terabytes per platter on an 8-disc configuration.

Okay. Yeah, so 10-ish terabytes per platter. That's, again our our current uh projections um and again i i'm not i don't want to diminish the amount of innovation that might be required between now and then but that looks to be right that's you know yeah that like like any roadmap that goes out that far that's the and if we can solve the problems that we see that look like they're solvable now. There was the other thing that came out. There was this writability limit with PMRs that exist today.

And it limited it to roughly a terabit per square inch. Is that? I mean, we're reaching that limit again. I think that the writability limit on PMR has probably moved up several times over the years. But yeah, somewhere between one terabit per square inch and maybe aerial density at, you know, sort of organically through heads of media, maybe six or seven percent a year, you know, as an industry. But, you know, we look at, we see that slowing down. And again, I hesitate to say it couldn't grow at all. But, you know, we probably looking at at something in the

lower single digit growth and so our projections are it would it would probably take you um you know till yeah that that leads you five or six years to the hard drives are too expensive to be buying yeah exactly so that so that you, going back to our earlier discussion, you know, you've got this relationship to NAND that you've got to contend with. And if that ratio in terms of cost per terabyte starts to squeeze, then you could see, you know, maybe more workloads begin to shift over.

So if you can maintain a kind of a 10x advantage, then, you know, the vast majority of the workloads, general purpose workloads, are still going to make more sense on hard drives. So that's really how we see it. If we had stayed on PMR, you know, I think that we see significantly more erosion to NAND from a market perspective, but by keeping sort of a 15% annual growth rate, which is what we are projecting for MAMR, we see the ability to maintain that advantage over NAND for, again, the big data applications, if you will.

And in terms of performance, minor performance increases as bit density leads to more bits per revolution, but nothing... Yeah, nothing, but it's kind of interesting. We've been looking at workloads a lot because, again, one of the issues, and you'd mentioned, why don't we go to taller drives? I mean, there is always this sort of IO density, so the IOs per terabyte that we have to pay attention to. But I think as we're looking at workloads, we're realizing the bottleneck in many cases, an increasing number of cases, is really throughput.

And if you look at throughput, NAND versus HDD, you don't see nearly as much benefit. Yeah, the gap is not nearly as large. It's not nearly as large. So again, then the HDD begins to make more sense. But trust me, we see NAND and HDD as definitely coexisting, both very important. But I think that unlike maybe the previous thought process, you know, we didn't see this sort of light switch over to NAND, you know, taking over the world. We think that the HDD is going to have a nice long life as long as we can continue to grow

density at this rate. Yeah, I think Ray and I have both been around long enough that we disbelieve anybody who declares some technology dead. Yeah. It was surprising to me how quickly the enterprise performance

sides of things started to switch over to NAND, really within the last couple of years because of the price differential wasn't there. And the performance was so much better. Yeah, I mean, it reached the point three or four years ago where SSDs and 15K RPM drives drives when you figured in all the costs were the same cost yeah but again that was that was a workload that was uh heavily biased towards ios right random ios right and i i think you know one of the changes we as storage architects have to make over the the next few years is to start, stop thinking about capacity and performance as independent and start

thinking about IO density. Right. And that, that IO density is really what determines what's the appropriate medium for things. Yeah. That's a whole different game, Howard.

Yeah. Yeah. We're old. We can change. Yeah. Old dogs,'re old. We can change. Yeah. Old dogs, new tricks, no problem. Let's hope.

So, yeah, you talk about big data and fast data in the discussion of, you're really talking about, you know, data analytics, data lakes, H know what to do with right now. And so I think that what we're finding is there are more efficient ways to store more of it. We're figuring out ways to extract more value out of it, which is sort of then this virtuous cycle that says, well, we need to store more because the more we store, the more value we can, we can extract. So yes. CEO walks into your office and says, I read this article in Forbes that said that data is like oil. And it's the most valuable thing we do. Never delete anything ever again.

There you go. Exactly. Our kinds of guys, we can hope we can only hope. Yeah, yeah, yeah, yeah. exactly we can hope we can only hope yeah but I mean but disk drives are also perfectly appropriate for rich data yes yes video surveillance

and IOT and all of these you know where big means not the aggregation but each object is big right and therefore throughput is what matters where big means not the aggregation, but each object is big. Right. And therefore, throughput matters really much more than IOPS. And to that, you know, on that point, that is what we're seeing is that the actual sort of block,

average block size transferred is growing as the data gets richer, right? So, you know, that's how we're starting to see workloads really shift in the cloud. Yeah, well, yeah, an Oracle database is one thing. An Oracle database where the main table has a blob attached to every record is another thing. So, I mean, you're actually using statistics from the field, that the block size being transferred on your typical disk is actually increasing year over year? Yes. Gosh.

That's an interesting statement. And I understand why it would be happening from a video and surveillance perspective, but there's more to this than meets the eye. Yeah, that's really a statement as to how, in particular, cloud architectures are changing. Yeah. Right. That's what we're seeing. So we're really analyzing, again, how our customers are using our drives to better understand what are the most important problems to solve.

Yes, definitely. And that is something we're noticing pretty much across the board. And it is really going to begin to inform future product direction. And Ray, if you think about it at the device level, just the widespread adoption of log-based data layouts would mean the drives are seeing bigger writes. Yeah. Yeah.

Well, it's not necessarily guaranteed that everything is log-based. No, but more and more. Certainly more. Yeah, yeah. As they get more and more richer data services and stuff. Well, and as storage systems that were designed to treat Flash well have logs to treat flash well, it rolls down here. I was going to ask the other question, whether the increasing block size is also apparent on flash as well as disk. But I don't know if you can answer that, Lenny.

Do you know? I can't answer that specifically, although we are, again, collaborating with the other side of the house to understand that. No doubt. No doubt. No doubt. No doubt. Not fair asking the district. Well, they're all the same company.

A few weeks from now, I might have the answer. But in fact, that's one of the things that we're going to be looking at. Okay, so if all the stars align properly, when will we see the first MAMR-based drives? And what are we talking about, a 20-terabyte drive by that point? Well, we announced that we're going to be launching production drives in 2019, which means our customers are going to see samples by the middle of 2018. We're not today announcing what the capacities will be. So I think that, you know. Yeah, but we're talking 30 months from now or so.

And you're at 14 today. So if you weren't already pushing 20, we would be going, hard drives haven't gotten bigger in a quarter. There was an announcement of, I think it was a 40 terabyte drive in 2025. That's what we're projecting. We could do on our existing platform, assuming that again, we,

you know, we, we can't necessarily predict that every generation of drives, we'll see exactly the same level of, um, you know, aerial density. Every once in a while,

instead of getting two to one, you only get 1.6 to one. Everybody goes, Oh yeah, it'll, it'll, it'll be probably a little uneven. But, you know, again, we've got some roadmaps not only for our product line, but also the actual MAMR technology. So we've already got two sort of enhancements planned beyond the first generation product. But again, we just, you know, we didn't make a definitive statement about what those capacities would be.

So I'm not at liberty to speak. And Howard, that 14 terabyte drive was SMR, so. Yeah, 14 terabyte drive is SMR. And I will say that before the MAMR drives come out, we have another PMR generation that we will be launching next year. So you can kind of connect the dots. Right. So what does this mean for BitPattern Media that we get to postpone it another few years? Well, I mean, bit pattern media is, again, not inherently incompatible with any of the things we've discussed.

So it can sort of develop in parallel. I would say that, you know, we, again, see enough growth in MAMR, you you know at this point that we feel that it it fulfills the the sort of near to midterm requirements uh we continue to invest you know again in in hammer today and other uh oh you know you're i mean obviously you have to hedge your bets with any technology that that's going to give you the next jump in capacity you know especially as the hard drive market has you know developed to be you know you and seagate and toshiba nipping at your heels you know i'm i'm frankly at the and why does toshiba stay in the hard drive business it's got to involve a lot of capital investment for the number of drives they produce.

I don't expect you to answer that question. I'm not. I'm not going to touch that one. Yeah, that's good. That's good. I thought you were smarter than that. So we've seen some other interesting things in hard drives over the years

that seem to have faded off, like intelligent object store hard drives. I know Western Digital had a project there a few years ago that ran a little Linux kernel in the drive. Not the only one, obviously. Well, Seagate had the Kinetic product, but we haven't heard much about either one of them in years. And more we've been seeing third parties doing interposer kind of things.

Yeah, the Starwind guys. Hybrid drives seem to have faded off the face of the earth. So is there anything for us to expect from them? You say hybrid, you mean SSD and disk in the same? Well, while trying to keep the 15K RPM drive competitive, some vendors said, what if we put like 8 gig of flash in there? What happens? As a cache, yeah. Lenny, is there anything exciting in the hard drive world

other than they'll be bigger next year? Well, I mean, again, I think the short answer is yes. But I'm- The long answer is you can't tell us about it. I can't really tell you right now. But I mean, I think it's fair to say, again, we continue to innovate on several dimensions, not just aerial density.

The aerial density is sort of the foundation for everything else though. And because we had chosen a path that maybe wasn't obvious to everyone, we felt like it was important to kind of come out and make that statement. But you will see. Yeah, it would have been really interesting if a year from now you said, and here's our new drive and it uses MAMR and you've never heard of it before. Yeah, exactly.

Exactly. So yeah. it uses mammer and we and you've never heard yeah exactly exactly um so yeah this well and and again because because we were not maybe as committed to um sort of a hammer at any cost uh path we were pretty silent and i think we felt like it was time to publicly state what we were working on. And again, we saw enough gain in terms of what we were, you know, seeing in the lab that, I mean, we've actually committed that 2019 product to MAMR. We've had that committed for some time. We just felt like now was the time to be public about it. But, you know, coming back to the, is there anything else exciting? I mean, I think that, you know, one of the things

that we're realizing as we look at, you know, what, again, what storage problems are we really trying to solve? You know, I think that increasingly the problems may be solved more at the platform level than just the device level. So we have to continue to extend, you know, leadership you look at software and hardware and the system and the application and, again, how the data needs to move through the system, I mean, there may be some real opportunities to sort of consider the storage, the unit of storage more at the shelf than the individual device. And I think that you can anticipate some things from us along those lines as we begin to leverage the device technology that we have on both sides, you know, HED with their benefits and NAND-based products with their benefits. And you can kind of imagine um you know some platform level solutions that might you know really leverage the best best of of each individual technology so you know not to pre-announce anything but i think i would be looking for those kinds of things from us yeah

when you when you start when you start talking about platforms, it opens an interesting dynamic in the storage industry because both Western Digital and Seagate have moved more dramatically into the platform business than they would have several years ago. I lost count of the number of times that disk drive vendors have come out with a bigger platform and then been scared out of that business by their OEM partners. But that doesn't seem to be happening the way it used to now that the hard drive business has consolidated. Well, I mean, there's that and there's also just, again, what is the nature of the problem we're trying to solve?

And you kind of almost have to move up a level to say, okay, is there a more efficient way to solve this problem than just producing a bigger, faster, cheaper drive? Right. And so I think that or or even you know how do i deal with the drawbacks of you know the the performance problems that the capacity creates yes and and if we start looking at shingle drives we say and we're going to build a file system where the log only writes in a full shingle at a time well then all of a sudden that problem goes away yeah well i i think i think too is and you talked early earlier in the conversation sort of about reaching the subatomic uh you know limits i mean you know well i mean when you start dealing with microwaves elevate you know

manipulating the electron spin torque in order to change the magnetic moment you've opened the door exactly but but my point is is that you know again at in a platform we have control over more of the variables and you know if you think about you know a hard drive and just you know interacting with other drives in a box you know vibration other effects i mean um you know if we if we have more control over the environment then we can actually um sort of stretch stretch in other dimensions so again i i just you know nothing specific to announce but but certainly right it's funny you talking that way um over the weekend i felt myself driven to write a series of blog posts once again explaining raid um and as i started thinking about it you know raid levels two and three which

just don't make any sense when you have disk drives that have logical block addressing um and then you know where we've placed our abstractions changes things and having a storage system that understands how the devices work below the first level of abstraction could come in really handy too. Yeah. So you mentioned vibration. Is there any rotational vibration changes with the MAMR technology? I mean, obviously, as the bits get smaller, I was wondering whether vibrations become more of a concern. No. I mean, again, we're continuing to sort of advance the mechanical platform,

the helium platform, the microactuation. And basically, we believe we understand all the factors and can continue to scale that side of the system to keep pace with what we're going to be doing on the aerial density side. So if you can get the bits to be small enough and stay where they're supposed to stay, you can manage to read them and write them. Exactly. Something like that.

There was quite a lot in the presentation about the intellectual property position that Western Digital has with respect to these technologies. I mean, quite an impressive number of patents in both Hammer and MAMR and that sort of thing. Is this prohibiting your competition from going down this path?

Or is there... I would say the industry has a rich history of cross-licensing. I couldn't imagine at this point with two major vendors that either one of you could operate without cross- patents well it's right yeah i i feel like we're i think we're all going to get to the same places eventually and and again i think that the reason we the reason we include that in our presentation was a couple fold one to to demonstrate that yeah we do actually have some pretty deep IP. And incidentally, we actually have more patents for Hammer than we had MAMR. So it's not like we made a MAMR decision because we had no idea how to do Hammer. I mean, it really was a decision based in a deep understanding of the competing technologies.

Right, right, right. Well, we reached our limit of time. Howard, is there any final questions you have? No, I think we've got it. Okay. I'm going to have to amend my post-apocalyptic statement, though. Let's hope. Well, I've been saying for a while that after the zombie apocalypse, there are going to be cockroaches, mainframes, and tape. And I think there's still going to be hard drives, too.

Okay, you can add us to the list. Great. Yeah, that's good. Okay. Lenny, is there anything you'd like to say to our listening audience? No, just thank you for listening. And again, I appreciate the support for the company and the interest in our technology. And we look forward to serving you well in the future.

Okay. Well, this has been great. Lenny, thanks very much for being on our show today. Next month, we'll talk to another system storage technology person. So, any questions that you want us to ask, please let us know. That's it for now. Bye, Howard. Bye, Ray.

Until next time