The PlayStation 5 and Xbox Series X/S consoles might have brought solid-state drives to more homes, essentially democratising what may well be called ‘next-gen’ in storage technology, but the quintessential hard drive isn’t going anywhere, not anytime soon anyway. Even as SSDs continue to grow and support the use cases where they make sense, companies like Seagate are also working simultaneously to unlock the next big thing in HDDs.
The last decade or so was all about increasing capacity—to the extent that Seagate is shipping 20 Terabyte drives today.
“The next decade will be about ramping performance,” John Morris, Senior Vice President and Chief Technology Officer, Seagate tells Financial Express Online, adding “there is already a big push to introduce enhanced performance capability in hard disk drives.”
According to Morris, Seagate is seeing a fairly high 40 plus percent growth in data demand associated with mass capacity use cases. The dominant storage solution in an Exabyte-scale data centre is—still—a hard disk drive. Typically, you would see at least 90 percent of the Bytes stored on hard disks in these large data centres and less than 10 percent stored on SSD. That ratio, he says, is in a rough equilibrium and it would continue into the next decade.
Generally speaking, while SSDs may gather a lot of limelight and see wide-scale adoption especially in the consumer space, the importance of a hard drive can’t be stressed enough. What’s interesting is, how Seagate plans to make them better.
FE: Could you take us through your current product and services portfolio—what is it that each of them brings to the table.
Morris: Seagate is a company focused on data, providing a portfolio of solutions to help people and companies manage data as efficiently as possible. We have a mix of different types of products. Our HDD portfolio is one of the broadest in the industry. We have a full family of products for both consumer and enterprise use cases. We have 2.5-inch and 3.5-inch hard disks for both enterprise and consumer applications. Our largest segment in HDD business is the mass capacity segment where efficient storage of large volumes of data is the primary objective. In SSD, we have SaaS and NVMe SSD and we’re really focused on enterprise applications. It’s not a large part of our business today, but it is an important part.
The bulk of our systems business is focused on solving enterprise pain points and we have a fairly robust engagement with our OEM partners to design and deliver these systems for their own use. These are storage centric systems and we also provide management software for that.
Lastly, we have an internal object storage platform that we are working on and that we have open sourced. Its purpose is to provide a highly scalable object storage software stack that can be used to enable efficient solutions for mass capacity use cases. We’ve recently announced our Lyve cloud which is a storage as a service platform that’s in the process of rolling out now.
FE: How has HDD technology evolved over the years? What are some of the latest advancements in the field right now? Can it be improved further?
Morris: There is no foreseeable end in the future of hard drives. We continue to innovate in recording technology and there is quite a strong conviction broadly in the industry that we have a technology transition to pick up from—the existing— Perpendicular Magnetic Recording (PMR) technology. We refer to it as Heat-Assisted Magnetic Recording (HAMR). When we and others go through that transition, we expect that to open up another 15 years.
If you look at the recording in the disk drive industry and the recording technology that we’ve been using, about every 15 years, there is a major technology transition. The last major technology transition we had was around 2005. That was when we transitioned from longitudinal magnetic recording and into perpendicular magnetic recording. Presently, we are all using perpendicular magnetic recording systems in hard disk drives. That’s given us in excess of 15 years of robust areal density growth. (*Areal density growth refers to how many bits per unit area you can record on a media.) We’re at a point where while there’s still ongoing improvement in areal density capability of perpendicular, we are reaching asymptotic limits. Probably none of us would be super comfortable to draw a line in the sand and we do keep surprising ourselves with innovations that continue to push the perceived limits forward, but somewhere in the range of 20 Terabyte, you’re going to start reaching diminishing returns with perpendicular recording.
We will be launching HAMR soon in the next couple of years. That would represent another major technology transition that we expect will continue to provide robust areal density growth for at least another 15 years. But that doesn’t mean that that’s going to be the last innovation. I think we will continue to innovate in this space and in another 10 years from now, it’s very likely that we will have another idea for a recording technology that can take over from HAMR, but the main point is, we believe that we’ve got 15 years of robust and sustainable areal density growth—that will be associated with HAMR—that we will productise in the near future.
We’ve also launched a dual actuator platform, which essentially doubles hard drive performance to address the desire for increased performance with hard drives. It’s a segment that is growing and we expect it will continue to grow, especially as we start looking at the kind of capacity per drive that can be achieved with HAMR technology.
FE: What was the need for an SSD then? Are they really better?
Morris: There are many different layers in the memory storage stack, and each has very specific strengths. For hard disks, the strength is their economic value. The cost per unit storage with hard drives is extremely good and it has a balanced performance capability. If you need that economic value, you’re going to gravitate towards a hard disk. SSDs—using flash media—shift the metric over to be a lot more focused on intrinsic performance capability.
If you look at hard disks, they are very good at streaming performance, so if you’re doing large file transfer or if you’re doing large block random IO, you can hit the peak performance capability of the hard drive with those types of workloads. But, as you gravitate towards smaller transfer length random IO, your performance starts to go down lower and lower. This is where SSDs really shine. Generally speaking, SSDs are capable of maintaining a flat performance irrespective of the workload that’s presented to them. (*That’s not completely true, but I think it’s close enough to the truth to be a useful example.) So, whether you present a small block random or a large block random or a sequential workload or a mixed write and read workload to an SSD, it’s going to have fairly constant performance. That’s a critical and intrinsic attribute of SSDs and it’s what allowed them to carve out a significant and very important role in the overall storage landscape.
You tend to use SSDs to solve performance challenges in your storage solution and you tend to use hard disks to solve cost challenges. And interestingly, when you look at very large-scale storage use cases—think, these Exabyte scale data centres—you have to use both. You need the hard drives to provide that economic value for the bulk of the data and you use SSDs to provide better service or better performance to access that bulk data through data tearing and data caching.
FE: Will SSDs get cheaper anytime soon?
Morris: SSDs use flash media and that flash media has a cost structure—that is associated with some architectural attributes of the flash itself. It is true that as you get more volume, you can achieve a level of scale, but you know flash in general is already at a level of scale where incremental volume is not a significant contributor to the cost profile of the flash itself. It’s really the technology. The principal driver for reducing the cost of flash currently is increasing the number of layers or decks that are in the flash media design and increasing the number of bits per cell and maybe, shrinking the planer dimensions of that cell. So, fundamentally, the cost of flash is driven by improvements made to those core attributes. They have an intrinsic profile for improvement that’s fairly well published in the industry.
We do expect to see continued decreases in the cost of flash. Likewise, we expect to see continued decrease in the cost of disk and generally speaking, the annualised improvements in the cost structure of both flash and hard disk are pretty close to one another and so our expectation is there’s kind of a rough equilibrium in that cost structure. In other words, the ratio of cost between a hard disk and SSD is going to be fairly constant over the next five or ten years.
FE: In HDD versus SSD debate, what’s the current state of privacy like?
Morris: Data security is fairly agnostic whether you’re using SSD or HDD. There’s no technical reason why you would have a different level of security for data stored on an SSD versus data stored on a hard disk. We have been providing security solutions for our storage products for quite some time. We adhere to international standards for the security technology and so there are standards associated with how encryption is done, and how the product handles key management, so on and so forth, and you know all of our security features on our storage products adhere to these standards to just make it easier for people to integrate and use the storage device with the security features and in their application.
FE: Given that you’ve partnered with both Sony and Microsoft to build custom storage for the PS5 and Xbox Series X/S, how do you look at proprietary storage?
Morris: We have a long history of providing OEM solutions for proprietary systems. Now, it’s also moved into the console space. If you look at consoles in particular, they have fairly rigid requirements associated with the performance matching between the external storage that you add to the system compared to what’s inside the box, the factory storage solution if you will. That performance characteristic of the storage is critical to ensure the end user experiences is seamless and high-quality and so there’s a desire to ensure adequate compatibility. Different companies will take different paths to ensure that you have adequate compatibility to provide the end user experience that they’re looking for and that will result in different solutions being created for the market. But at the end of the day, the objective is to make sure that the storage is providing a seamless end user experience.
FE: Is there a third technology also simultaneously in the works—aside from HDD and SSD—that the industry is currently exploring?
Morris: Generally speaking, for storage use cases, there’s really nothing in the horizon that looks like it has an opportunity to disrupt either SSD or HDD, in their respective areas. Certainly not in the next 10 years. There are technologies out there that could mature and could potentially offer opportunities to participate in say, archive storage, but we’re really not seeing any disruptions in the horizon for at least another 10 years relative to where HDD and SSD are operating today.
FE: How immune is Seagate to the global semiconductor shortage? Has the pandemic boosted overall sales?
Morris: I don’t think anybody is really immune to it. There are clearly supply chain challenges in the electronics space. We have good agreements and partnerships with our suppliers in the electronics supply chain, and so we’ve been able to maintain our production capability. We expect that capability will persist. Everyone is keeping a close eye on everything that’s going on here, but you know, I would say, largely speaking, our production capability has continued to be robust despite the challenges that are occurring right now.
The pandemic did accelerate the adoption of tools and solutions to enhance edge infrastructure capability as well as endpoint capability and so, there is an acceleration of deployment of solutions. But as far as demand, we’ve had a pretty stable historic demand even through the pandemic, and we expect those demand profiles will continue.