by Anton Shilov
07/11/2012 | 03:25 PM
In the world of widespread digital content and documentation, storage sub-systems become tremendously important. But with demands for different qualities of storage - performance, capacity, reliability, portability, etc., it is impossible to distinguish, which of the modern storage technologies - hard disk drives or solid state drives, local or network-based - are the most important for today and for tomorrow and while will eventually prevail. In fact, now it looks like both solid-state and rotating media-based storage devices will continue to prosper for many years to come. In a bid to find out, what is going on with HDDs, SSDs now and what is likely to happen in the coming years, we decided to talk to a storage specialist from Toshiba, the only major company that sells all types of storage devices for different segments of the market.
X-bit labs: X-bit labs: Hello, thanks for the opportunity to discuss the forthcoming generations of storage devices. Please introduce yourself.
Joel Hagberg: Joel Hagberg, vice president of marketing at Toshiba America Electronic Components, Storage Products Business Unit.
X-bit labs: Many HDD makers believe that only hard drives can satisfy demand for storage capacities of the world as NAND flash fabs are too expensive to build. Do you share this belief?
Joel Hagberg: The production of semiconductor and NAND Flash components is certainly a capital-intensive endeavor, but the design and manufacturing of rotating storage solutions is as well. The cost and expertise involved in producing high-quality hard drives in high volumes is one of the fundamental reasons behind consolidation of the HDD supplier base. Toshiba’s view is that both HDDs and SSDs play important roles in meeting the demand for housing the exponentially-increasing amounts of data that is created and shared every day. We see HDD and SSDs working synergistically in a storage ecosystem where one technology no longer suits all applications. Toshiba’s storage portfolio strongly reflects this view of storage today, and points to where we are heading in the future.
X-bit labs: Will solid-state storage possibly ever be able to substitute rotating media?
Joel Hagberg: The future is not about substitution of storage, it is about strategic deployment. Global enterprise needs the right storage at the right place at the right time. Information and data are growing both in volume and in value. We are really just at the beginning of the software, hardware and services that will reshape data centers and enable them to tackle the mammoth amount of big data that is created every day. Cloud computing requires tiered storage to help tackle the information deluge, and to enable companies to use that information in the most optimal way. HDDs and SSDs are both critical to these architectures, and will power the enterprise moving forward.
X-bit labs: Do you think that going forward - perhaps, three to five years from now - different applications will use different storage technologies? E.g., automobiles will use solid-state drives, local storage systems for entertainment content will stick to hard drives, and enterprise will use both for different kinds of usage and so on?
Joel Hagberg: It depends on what provides the most value to the application. Tablets use solid state memory to optimize for speed, power consumption and robustness; desktop PCs use HDDs to optimize for capacity and cost. Similarly, enterprises which consist of many applications will use strategically architected combinations of HDDs and SSDs. As an example, the need for speed in stock market systems is very clear and provides enough value for the company to afford to use faster storage – traditional enterprise HDDs and high performance enterprise SSDs. Email archive may be less urgent, and can be properly supported by nearline drives. A continuing evolution and specialization on a by-application basis is the future of storage deployment.
X-bit labs: Can you estimate how many exabytes of NAND flash and magnetic media will be made this year, next year, etc?
Joel Hagberg: Toshiba does not provide public production forecasts. Analyst firms such as IDC and Gartner are better sources for that type of data.
X-bit labs: Would you expect cloud storage to lower importance of local storage capacities? Perhaps, it can kill various USB flash sticks completely...
Joel Hagberg: The debate over localized storage versus cloud-based storage is dependent on many factors, including the cost per gigabyte of storage, the amount of storage capacity involved, the speed and pervasiveness of available connectivity, as well as the time management aspect of personal data and where it may reside. People still have valuable uses for USB flash drives and external drives. The cloud is a reliable data backup option, but the need for personal storage that can easily be touched is something that the average user will not give up easily.
X-bit labs: What do you think are the primary challenges for the SSD industry in general? Aren't the vast majority of early problems (which occurred in 2008 - 2010 timeframe) solved and consumers can safely install solid-state drives instead of hard disk drives into their PCs?
Joel Hagberg: The primary challenge for the SSD industry is to balance the cost per GB and IOPS vs. traditional HDD storage costs. SSDs are a reliable storage technology that consumers and enterprises can, and do, rely on for their storage needs every day. They provide the ultimate performance for those applications with a higher budget for storage and less need for capacity. The cost of SSDs has come down significantly, but HDDs are the best fit for higher capacity, lower cost storage applications.
X-bit labs: Will there be any breakthroughs in terms of SSD affordability this year? Perhaps, in terms of capacities?
Joel Hagberg: SSDs will continue to advance to higher density in smaller lithographies – essentially providing more capacity in the same package size generation over generation. The technology process offers this benefit, but the affordability is also dependent on the application’s needs as well as the supply and demand process; for this reason it is difficult to predict. However, it is clear that an increasing number of applications see the value and there will be consistent, healthy growth in SSDs.
X-bit labs: Some believe that SSDs and NAND flash are complementary storage technologies, not primary. If this is so, do we need breakthroughs at all?
Joel Hagberg: SSDs and NAND flash are both used as primary storage methods today. SSDs are the primary storage in Ultrabook/ultrathin laptops and enterprise tiered storage systems. NAND is the primary storage medium in smart phones and MP3 music players. SSDs are complementary to HDDs: they can augment responsiveness and performance characteristics of larger capacity storage devices, they can serve as small capacity, standalone storage in devices that connect to other storage entities such as the cloud, and they can be combined into a single entity that offers both massive storage capacity as well as ultra-fast responsiveness. With the demand for faster data delivery on websites and servers, there is always a need to advance technology to meet such needs. Toshiba sees opportunity in the storage as it stands today, as well as in emerging, innovative solutions; we don’t see any slowdown in the long history of innovation that has characterized the storage industry.
X-bit labs: Do you expect ultrabooks to give SSDs an opportunity to become mainstream in mobile computers? Some expect share of ultrabooks to increase to 43% of all laptops in 2015, pretty sweet piece of the market, isn't?
Joel Hagberg: New processors, new operating systems, new industrial designs and a greater range of storage options are all adding value and a renewed experience for PC users in the consumer and commercial markets. Thin and light notebooks are popular because users are always on the go and need the flexibility of such mobile devices; they are a premium product and will continue to gain acceptance in the market with consumers who value sleek and slim devices. SSDs will be a high-performance option that will be a good fit in the higher storage budget systems, but there will be significant volume of Ultrabook/ultrathin systems that ship with a thin HDD for lower overall systems cost. Many Ultrabooks will also ship with a dual drive configuration combining a hard drive with a low capacity SSD module.
X-bit labs: What, in your opinion, will help to drive SSDs into desktops?
Joel Hagberg: The current environment is ripe for innovation and innovative approaches. Toshiba’s value is to deliver the storage technologies that can transform those concepts into reality. Depending on the system design, SSDs can be used to boost responsiveness and performance attributes. Desktop systems have traditionally focused on cost reduction, which has not been conducive for adoption of SSDs. Recent innovation in all-in-one flat screen systems and the potential for instant response with new operating systems may enable increase penetration of SSDs into desktop systems.
X-bit labs: Will you develop your own controllers for own SSDs eventually, or will you continue to rely on third party controllers and in-house firmware?
Joel Hagberg: Toshiba is always looking to integrate the best controller technology into its SSDs. The company uses its own controller design in a range of the current Toshiba SSD offerings today. We are open to both internal and external controllers to maximize performance, reliability, time to market and cost effectiveness.
X-bit labs: Currently you use SLC NAND flash for enterprise SSDs. I wonder whether you plan to develop special versions of multi-level cell NAND flash memory with better endurance for your enterprise solid-state offerings.
Joel Hagberg: There is a requirement for both types of NAND in the enterprise segment. Depending on workload requirements and application types, one type of NAND may be a better fit compared to another type. Using enterprise grade eMLC NAND also helps with the cost metric, and provides higher capacities. In the end, it depends on the write endurance you need.
X-bit labs: Are there any chances that developers of operating systems will address wear leveling issues of SSDs? Or will optimize OSs for SSDs by other means? When is this expected to happen?
Joel Hagberg: That is a question better answered by the developers of operating systems.
X-bit labs: Are there chances that the market of SSDs will commoditize any time soon
Joel Hagberg: Industry analysts firms such as Gartner and IDC are better sources for answering this question.
X-bit labs: But maybe you have your own opinion? Are there price wars on the market of SSDs? Have you felt the price pressure from your competitors in the retail or other (OEM, channel, etc.) markets?
Joel Hagberg: The NAND memory community is continuing to develop new technology which will improve capacity and lower cost per GB in each lithography change. As we move to SSDs based upon 1x nm NAND, we will enable higher capacity and lower cost SSDs. However, there are significant costs required to increase NAND FAB capacity and technical challenges must be overcome to make lithography changes. Because of this, there will continue to be a significant difference between the cost per GB of SSDs and HDDs. In conclusion, even though SSD prices will continue to drop over time, they will not approach the commodity storage range of hard disk drives for a very long time.
X-bit labs: Do you plan to start using three-bits-per-cell (3bpc) NAND flash memory in your solid-state drives this year?
Joel Hagberg: Three-bits-per-cell is an interesting technology in which we are deeply involved. Our engineers are always innovating and researching future storage technologies. At this time, we have nothing specific to announce, but this is one of the future NAND flash technologies that look promising.
X-bit labs: Well, your manufacturing partner, SanDisk Corp., openly said that TLC/3bpc is viable for consumer SSDs, SandForce has created SF-2000 controller that supports TLC/3bpc NAND flash made by Toshiba. So, what are your expectations? Will you be ready with consumer SSDs featuring the 3bpc NAND this year or early next, or, maybe this fiscal year?Joel Hagberg: As I commented above, clearly there are a range of new technologies that look promising for future designs but we do not comment on future product designs and do not have any SSD announcements to make at this time.
X-bit labs: Do you intend to use two-bits-per-cell (2bpc) MLC NAND flash memory made using 10nm-class [19nm, to be precise] process technology inside your SSDs in 2012?
Joel Hagberg: Our engineers are always innovating and researching future storage technologies. We have nothing to announce at this time regarding future NAND flash technology in SSDs.
X-bit labs: So, what are the primary applications for the MLC NAND flash you make using 19nm fabrication process?
Joel Hagberg: The primary application for our MLC NAND is the full range of consumer products enabled by NAND today, including USB Flash Memory Sticks, MP3 players, SmartPhones, Tablets, eReaders and SSDs. We announced our new range of 19nm based SSDs, the THNSNF-series on June 4, 2012; you can the full news here.
X-bit labs: Do you expect new types of non-volatile memory like memristors, PRAM, MRAM, etc., to become commercially viable within the next three to four years?
Joel Hagberg: Toshiba does not provide product forecasts. Industry analyst firms such as Gartner and IDC are better sources for answering this question.
X-bit labs: Maybe you can name a technology or two that you are working on now which may eventually become a new type of non-volatile memory? Perhaps, in longer term...
Joel Hagberg: There are a number of new memory technologies that Toshiba is working on for future storage applications. There are new NAND lithography steps that look promising beyond 19nm but we expect these to remain in the teens for the next two years. Toshiba is also investigating three-dimensional storage as a potential successor to NAND Flash Technology and we expect to have prototype samples in 2013. MRAM also looks like it may have some interesting applications in the coming years as well.
X-bit labs: Just like others, you suffered significantly from the Thai flooding last year. Can we say that by now you have recovered all your capacities?
Joel Hagberg: Yes, Toshiba successfully ramped up additional volume at its manufacturing facilities in other countries following the Thailand floods. As part of our asset exchange with WD, the former Toshiba disk drive manufacturing facilities in Thailand are now owned by WD.
X-bit labs: What can you tell about the supply chain? Are they back with pre-flood levels?
Joel Hagberg: The entire HDD industry was impacted by the supply chain recovery process following the Thailand flooding. Certain aspects of the channels are still recovering, but the overall supply chain is getting back on track. Industry analyst firms such as Gartner and IDC are better resources for more detailed comment.
X-bit labs: Now that you own Western Digital's 3.5" HDD production capacities/equipment, should we expect any tangible strategy shifts when it comes to hard drive business?
Joel Hagberg: Toshiba’s HDD business has continued to evolve and grow to meet market and customer needs. The past few years, Toshiba has expanded market coverage to include enterprise products with its strategic acquisition of Fujitsu’s storage business. Last year, it consolidated its HDD and SSD businesses under one group to provide customers with leverage and efficiency for everything from consumer to cloud. The addition of the 3.5” drives for desktop and consumer electronics (CE) is the latest in this series of expansions to support market changes and customer needs, ultimately meeting our strategic goal to double our revenue during the next few years.
X-bit labs: How will your HDD strategy change with the recent acquisition?
Joel Hagberg: We expect incremental growth resulting from the addition of the desktop business. The acquisition is another demonstration of Toshiba’s strategy to expand the scope and depth of our offerings as well as our commitment to storage technologies.
X-bit labs: While previously you did have a couple of 3.5" enterprise HDD in the product line, now that you own WD's manufacturing equipment, will you introduce consumer-oriented 3.5" hard drives, or continue to concentrate on 2.5" HDDs for consumers and carry a couple of enterprise 3.5" models?
Joel Hagberg: To Toshiba, 3.5” 15K enterprise is no longer a growth market; we are now fully supporting customers in the transition to small form factor (SFF) in servers and storage. Our transaction with Western Digital includes products and equipment (not facilities) and will enable a quick, cost effective ramp for both desktop and CE products. Our portfolio now includes the full range of what our customers need, including 3.5” desktop and CE drives, 2.5” mobile drives, 2.5” enterprise drives, 3.5” nearline (capacity optimized drives) as well as client and enterprise solid state drives.
X-bit labs: Toshiba is known for breakthrough 2.5" hard drives for laptops. But is not known for leading-performance or leading-capacity HDDs for desktops. Why is that?
Joel Hagberg: In the past, Toshiba has concentrated on development of small form factor mobile HDDs and enterprise HDDs. We have not participated in the desktop market segment until the recent acquisition of the HGST desktop design from WD. Toshiba is committed to both the rotating media and solid state segments of the storage industry. Over the years, we have built upon the solid foundation we established on 2.5- as well as 1.8-inch HDDs. In 2010, Toshiba acquired Fujitsu’s HDD business and added enterprise-class HDDs to its portfolio. Then in 2011, the Toshiba semiconductor and HDD businesses joined forces, resulting in the addition of enterprise and client SSDs to our portfolio. This year’s acquisition of WD’s 3.5-inch desktop HDD assets and IP launches us into the client 3.5-inch HDD space with models optimized for both the computing and consumer electronics markets. There is no other company in the industry that has the breath of products, storage technologies and IP ownership that Toshiba owns.
X-bit labs: Now that you own WD's manufacturing facilities/equipment, how many hard drives a quarter can you ship?
Joel Hagberg: Toshiba does not provide public forecasts of product shipments. However, as was previously announced, we obtained manufacturing equipment from WD, not an entire facility.
X-bit labs: Does it mean that unlike Seagate and Western Digital, Toshiba does not indicate how many drives a quarter it produces? For example, how many drives did you ship in calendar Q3 2011 and in calendar Q1 2012? What kind of increase, in percentages, perhaps, should we expect thanks to addition of WD's equipment?
Joel Hagberg: In 3CQ11 we shipped more than 21 million HDDs, and we were able to exceed that level with our existing product lines in 2CQ12. The addition of the desktop product lines will enable Toshiba to increase its capacity by 30% in late 2012.
X-bit labs: When do you expect that WD equipment to be installed and start operations?
Joel Hagberg: We expect to begin production in our facility in 3CQ12 and ramp to full capacity of the existing lines by the end of 2012.
X-bit labs: Do you think the hard drive industry is finished with consolidation and no further transactions are logical to expect?
Joel Hagberg: The hard drive industry is still in the consolidation process with both the Seagate-Samsung acquisition and the WD-HGST acquisition. It will be 1-2 more years until this process is final and the companies are fully integrated. The consolidation was driven by the need for increased R&D to advance areal density, and the challenge to achieve ROI as a result. Consolidation beyond this point is hard to predict.
X-bit labs: Do you expect 2.5" HDDs to pose threat to 3.5" drives, as some analysts recently said?
Joel Hagberg: The demise of the 3.5-inch form factor is greatly exaggerated. We see strong demand for capacity-optimized HDDs for enterprise applications as well as client HDDs in the desktop and consumer electronics spaces. The 3.5-inch form factor fits specific requirements and needs of these applications. Toshiba does not expect 2.5-inch drives erode the 3.5-inch market as there is need for both depending on the use case.
X-bit labs: Seagate predicts slower growth of HDD capacities because of shortages of hard drive components. Do you share this belief?
Joel Hagberg: The virtual vertical integration model I spoke of earlier encompasses strategic, long-standing relationships that we have built with critical component suppliers. Together, we develop and test new recording and storage technologies, share IP and collaborate on the build-out of these technologies to support volume production with consistent quality and reliability. Certainly the technology has gotten more difficult, causing the annual rate of storage density per disk to slow. However, Toshiba will continue to innovate and bring to market new products that encompass state-of-the-art technologies that we and our suppliers have developed. At this time we do not anticipate any near term component shortages in 2012.
X-bit labs: Will this process cause a per-HDD/per-GB price increase? Or rather force hard drive makers to find new ways to make HDDs, perhaps, by further unifying different drives?
Joel Hagberg: HDD pricing, like most pricing, is mostly the result of supply/demand dynamics. In addition, serving cost-conscious applications in both the consumer and commercial HDD industries has always been driven to maximize efficiency through technology and operations. Any and all avenues of cost improvement that produce high quality reliable products are continuously sought after and explored.
X-bit labs: Both Seagate and WD have their own magnetic media production. You continue to buy media from other companies. Maybe it makes sense to bring that in house?
Joel Hagberg: We are virtually, vertically integrated with our long-term partners - TDK and Showa Denko KK. We continue to co-develop with them in our Advanced Technology centers. In addition, the entire HDD industry continues to rely on these vendors for industry leading areal density advancement. Our competitors have significantly larger employee head count requirements to support their component development and manufacturing requirements. Toshiba is able to operate our storage business with significant lower headcount expense with our virtual integration model.
X-bit labs: When do you expect to start to use heat-assisted magnetic recording (HAMR) technology in your HDDs?
Joel Hagberg: Toshiba is working independently as well as collaboratively with TDK and Showa Denko to develop next generation recording technologies. Thermal Assisted Magnetic Recording is just one candidate. We are also working with other energy-assist technologies as well as discrete track (DTR) and bit pattern media for the possibility of future integration into HDDs.
X-bit labs: Do you expect new HDD media players to emerge on the scene at the HAMR or post-HAMR era? Looks like the media production needs both R&D and capital, companies in that case will be able to stay independent...
Joel Hagberg: It is not altogether clear that HAMR will be the next major magnetic recording system to be commercialized. The HDD business is very capital intensive. We see the virtual vertical integration model as providing access to the best minds and technological expertise required to remain a state-of-the-art supplier on an ongoing and consistent basis. Toshiba feels that it will benefit from its collaborative development and investment model with the current component suppliers for advances in future heads and media design.
X-bit labs: Why do you think that HAMR may not be the next major magnetic recording technology? Seagate has already demonstrated technology and promised 1Tb/inch2 areal density. Meanwhile, DTR is essentially a [+50%] boost for PMR and it probably has been deployed in today's hard drives. Bit patterned media promises to offer 1Tb/inch2 - 3Tb/inch2 areal density, so it may be considered as a post-HAMR technology, unless, of course, you have found a way to implement it ahead of HAMR :). Is it so?
Joel Hagberg: The answer is both a statement regarding the term HAMR, which is a vendor’s proprietary term, and of manufacturing challenges. The industry may soon come to term the technology TAMR (Thermal Assist Magnetic Recording), and may or may not include lasers as microwave heating technology is also being explored as an option outside of the lasers that have been used in previous demonstrations of HAMR or TAMR. Toshiba is the one HDD vendor that has high volume production experience with lasers from its work with CD and DVD operations. We look at both the advantages and challenges associated with multiple lasers mounted in an enclosed HDD design and feel that there are options which should be explored beyond lasers.
X-bit labs: When do you think two dimensional magnetic recording (TDMR, 10Tb/inch2 areal density) becomes viable?
Joel Hagberg: There are a number of complex technology development and manufacturing issues which need to be addressed to approach the 10Tb/inch2 level areal density. It is difficult to speculate when the challenges may be overcome and enable such significant achievements in manufacturing technology.
X-bit labs: Will rapid growth of HDD's areal density improve performance of hard drives to levels of SSDs in foreseeable future? Or performance niche is now dominated by SSDs?
Joel Hagberg: Certainly increasing the areal density in HDDs has performance benefits, which alone cannot approach the level of the fastest SSD. Toshiba believes that HDD, SSD and NAND technologies can be used together to boost performance and deliver massive storage capacity for the user.
X-bit labs: Being one of the largest NAND flash maker in the world and with both SSDs and HDDs in the portfolio, isn't it logical for you to release hybrid disk drives that combine performance of SSDs with capacities of HDDs?
Joel Hagberg: Toshiba is very excited about the current storage technology environment. We see different approaches and innovations that will take advantage of both rotating media and solid state memory technologies. Because our storage technology portfolio includes NAND, HDD and SSD, Toshiba is very well positioned to service customer requirements across these storage technologies. Hybrid HDDs are another solution that Toshiba will add to its storage portfolio.
X-bit labs: When can we expect you to jump on the hybrid HHD bandwagon?
Joel Hagberg: Toshiba’s view on the Hybrid-HDD market is that while there is much promise to the hybrid concept, industry adoption has been fairly lackluster to date. Some of this may be due to early products not meeting market expectations and host applications not being able to take advantage of the technology. Toshiba’s approach is to launch new technologies when the market is ready to deploy and use that technology in ways that make a compelling difference to the user experience.
X-bit labs: Why do you think hybrids have not gained much traction on the storage market? Intel's caching solution also has not gained much success...
Joel Hagberg: There are a number of sophisticated elements that need to come together in order to realize the full potential of hybrid HDD technology in user applications. The level of responsiveness and performance that we would like to see associated with this upcoming generation of hybrid HDDs requires cooperation and collaboration from components in the ecosystem. Early caching solutions and hybrid drives have been waiting for the operating system providers to enable features that deliver performance value to the end customers.
X-bit labs: Do hybrid storage solutions actually wed performance of SSDs with capacities of HDDs? Or is it still a big compromise?
Joel Hagberg: Hybrid solutions increase the performance of the storage device in a manner that is still very cost effective for the consumer. I would not call this a compromise but rather one solution to improve I/O performance.
X-bit labs: There are two ways hybrid drives can be implemented. One way is to let NAND flash cache data from frequently used LBAs of hard drive, another way is to install plenty of NAND flash memory and simply install demanding apps onto the flash drive. Which one is better? Perhaps, there is a third way?
Joel Hagberg: This gets into the realm of technology integration and future design decisions. There are trade-offs involved, and a bigger amount of NAND Flash is not always the best way to achieve a desired result.
X-bit labs: Do hybrids make sense in enterprise or server environments?
Joel Hagberg: Toshiba continuously watches the market and see if a hybrid solution can benefit in the enterprise space. We do see some potential benefit for such a device, however, at this time Toshiba offers SSDs and HDDs for all required tiers in the enterprise segment. We do not see much demand for a hybrid enterprise HDD at this time.
X-bit labs: Thank you very much for the detailed answers, Joel!