The storage industry continues to evolve, with hardware systems maturing and software systems reaching adolescence. Following the storage-area network/network-attached storage (SAN/NAS) METAspectrum in February 2003, this report takes a "midterm" look at the various vendors in the context of new technologies and current strategies.
META Trend: Storage management automation, standards, and process will remain immature through 2006. Net annual storage growth will average 35%-40% for enterprise (monolithic), 45%-50% for midrange (modular), and 80%-85% for capacity-based (SATA/ATA). Like-for-like price/capacity will improve 35%/year. Through 2007, storage hardware will be rendered a tiered commodity by software-based information life-cycle functionality (e.g., storage resource management, data protection/recoverability, integration, data movement, interoperability), which will become the primary enterprise storage differentiators.
Storage management remains a dynamic and sometimes volatile marketplace. Although the days of extreme storage growth (e.g., 80%+ CAGR) remain a memory, storage demand has risen steadily during the past 12-18 months. Storage hardware has reached a high degree of maturity and substitutability, while storage management software continues to evolve. Several vendors now claim heterogeneous management capabilities, though the actual functionality often is quite limited. Similarly, vendors are lining up behind the Storage Management Initiative Specification (SMI-S), with many claiming “compatibility.” This is good news for storage users in the long term (and is an inevitable element of the maturation process), but we do not expect robust SMI-S implementations before 2006/07. Heterogeneous functionality will be delivered in relatively small increments during the intervening time.
The SAN/NAS METAspectrum includes graphical representations of relative vendor positioning based on two axes, presence and performance. Most technology and product advances since the chart was created have been incremental in nature, so a full re-evaluation is not justified at this time. However, many of the subsequent vendor announcements are worth noting. Therefore, the focus of this Practice is limited to two primary areas: technology and strategy. The discussion of each vendor considers enterprise SAN, midrange SAN, NAS, and storage management software to the extent that any individual company participates in those market segments. To avoid any appearance of imparting a leadership designation, the vendor discussions are in alphabetical order. It should be noted that Veritas is not included in this report because its ServPoint NAS product (using Sun hardware) has been discontinued.
The table in Figure 1 below identifies each vendor in each of these market segments using a positive (up arrow), neutral (sideways arrow) or negative (down arrow) designation chiefly to graphically depict whether or not a given vendor has both vision and execution in that specific marketplace. (Note: The arrow designations are relative to the specific vendor only. They are not relative indicators between vendors, do not indicate leadership, and do not imply substitutability between vendor product suites. For an indication of leadership and relative position, refer to the original METAspectrum.)
Although Dell does not develop storage technology per se, its entry into any market can change the market dynamics. Typically, Dell focuses on maturing product markets where its remarkable manufacturing and logistic efficiencies allow it to produce a reasonable product at the lowest possible price.
Dell’s partnership with EMC to sell the Dell/EMC Clariion product line was not Dell’s first attempt to partner its way into the storage market, but it has certainly proven to be its most successful. Early concerns about channel conflict between Dell and EMC have been proven to be unfounded. The partnership between the two companies is contractually bound through 2008.
In addition to Clariion, Dell offers its own PowerVault line of direct-attached storage (DAS), SAN, and NAS devices, aimed primarily at the entry-level and small and medium business (SMB) markets. The ascent of Advanced Technology Attachment (ATA)-based products would seem to be an ideal product category for Dell, and indeed, the company will be offering various new products in this arena. Dell’s PowerVault NAS products include both Microsoft Storage Server 2003 and EMC’s Clariion NS-series subsystems.
Dell’s primary contribution to the marketplace is in driving the adoption of standards. To the extent that standards are established, Dell can best exploit its logistical strengths. Moreover, standards implementations tend to make product deployments simpler, with a lower services component. As a result, Dell’s intellectual property contribution focuses on tools to enhance deployment of the technologies that it sells (e.g., Interactive Storage Explorer, Storage Advisor). In terms of storage software, Dell partners not only with EMC, but also with Veritas, CommVault, Legato, and others to deliver storage resource management, backup/recovery, and data movement software. For organizations that know what they want, and when Dell carries it, Dell is certainly a solid, proven supplier.
After dominating the storage industry in the 1990s, EMC found itself with an aging flagship product, a diminished software technology advantage, and an out-of-line cost model just as storage demand tanked in 2001/02. To the company’s credit, it has turned each of these weaknesses around, has once again regained thought leadership, and is now setting the agenda for the storage industry. EMC’s aggressive pursuit of industry heavyweight management has paid off by revitalizing the company’s technology and market initiatives. The company offers breadth and depth in enterprise-class systems, modular systems, ATA arrays, and storage management software.
Beginning with its introduction of Centera and CAS (content-addressable storage), EMC has changed the storage management industry by focusing on content rather than on mere device management. Its acquisition of Documentum gives it a strong position in content management, an area which is becoming more infrastructure-like in nature (see Flash 2008). Moreover, this dovetails with EMC’s information lifecycle management (ILM) initiative in which data is managed based on content. Although EMC did not invent the ILM process or the term, it has championed its way into a leadership position in this area. In addition, the company is building a professional services organization in order to fulfill the services portion of ILM.
From a product standpoint, EMC’s primary weakness is in NAS. The company offers Celerra on both Symmetrix and Clariion, the NS-series on Clariion, and the NetWin platform using Microsoft Storage Server 2003. Although EMC continues to struggle with usability of these systems, which are purchased primarily by dedicated EMC IT shops (and often as part of a bundled package), these systems remain the primary competitor to Network Appliance (NetApp) for enterprise-class NAS.
The departure of several key storage executives during the past 18 months has forced HP to regroup the business unit to some degree. However, HP’s storage business will carry on through sheer force of momentum, if nothing else, until the new management team can re-energize and redirect the group. HP systems remain shortlist candidates. The company offers one of the broadest portfolios of products in the industry, from the entry-level MSA (Modular Smart Array) series to the EVA (Enterprise Virtual Array) family of modular system and the monolithic XP arrays (OEMed from HDS). HP also offers a very strong line of storage management software that integrates with its OpenView management framework.
During the management transition, HP’s new product initiatives have been rather quiet. Under the prior management, HP completely missed the market movement toward ATA-based arrays, but it will offer serial ATA (SATA) drives in MSA arrays starting in May 2004. Even though most of the company’s product initiatives are incremental improvements on existing products, HP’s CASA storage virtualization products continue to be among the more innovative product offerings. HP’s storage management solutions include heterogeneous remote and local replication, business continuity, storage resource management, and much more. Recently, HP has added a secure content and archival solution (acquired from Persist Technologies) called Reference Information Storage System (RISS).
HP has also significantly cleaned up its NAS portfolio. Previously, HP offered both its own NAS solution as well as Microsoft-based systems in a bewildering array of platform/storage options. It has since narrowed its platform to the Microsoft Storage Server 2003. Although this makes differentiation from other competitors (e.g., Dell) more difficult, the change is still positive. HP’s NAS line appeals primarily to IT organizations that want the simplicity of an all-HP solution, since NAS has never been a primary focus of the company.
The current challenge for HP is to regain its vision and momentum in the industry. Although HP has no significant product gaps (other than an ATA array, as mentioned earlier), it has been relatively late to market in key growth segments such as backup appliances, content management, and ILM. Moreover, HP is betting its storage management future on SMI-S standards, which we believe inherently lag in the industry (see SIS Delta 1019).
Hitachi Data Systems
HDS’s Lightning 9900 V Series of monolithic storage arrays is the counterbalance and chief protagonist of EMC’s DMX product in very large-scale deployments. Indeed, based on client feedback, the 9900V is quite likely the most reliable system in the industry. IT organizations with very large-scale needs (e.g., >30TB, highly transaction-intensive applications, high availability) will quickly narrow their choices to the 9900V and DMX. Moreover, the 9900V is as comfortable in an IBM zSeries FICON environment as it is in a Unix/Windows Fibre Channel environment. The 9900 V series has been on the market since May 2002, and a product refresh can be expected later in 2004. The 9900 V series is sold primarily through HP (OEM) and Sun (reseller), though HDS does have a direct sales force and other value-added resellers.
HDS’s Thunder 9500 V Series has not enjoyed the success of its larger sibling. This lack of success is largely due to very limited sales channels. Given that HDS uses indirect channels as its primary go-to-market medium, the 9500 suffers because it currently has no major OEM support.
HDS’s NAS offerings historically have been weak. Within the past year, however, HDS has partnered with NetApp to offer the gFiler. The gFiler allows a portion of 9900 or 9500 storage arrays to be allocated as NAS storage. For HDS customers, this allows the use of a single storage subsystem for both SAN and NAS, driving higher utilization. Yet the go-to-market strategy is somewhat problematic. Neither HP nor Sun is likely to resell the gFiler, since it makes little sense to be an OEM of an OEM (though gFiler could be resold directly from NetApp). Therefore, the sales forces of HDS and NetApp often have crossed paths, resulting in some painful channel conflict; both companies claim to have solved the problem.
The primary weakness of HDS is storage management software. This is an area where the company tends to be more of an imitator than an innovator. Nevertheless, its HiCommand suite offers satisfactory functionality for the vast majority of requirements. Moreover, the company has partnered with AppIQ to deliver heterogeneous storage management capabilities. Sun has joined HDS in delivering this product line; HP has not and probably will not follow suit, due to the overlapping functionality with HP’s own products. HDS has also partnered with IXOS to provide e-mail archival solutions that leverage 9900 V Series LDEV Guard features to enable a compliant e-mail archival solution.
IBM’s F-series Enterprise Storage System (ESS, a.k.a. “Shark”) had a less than stellar market experience, as was reflected in the previous Enterprise SAN-Attached Storage: Market Overview. Since that report’s creation, IBM released its model 800 and 800 Turbo Shark systems. These systems have proven to be very reliable, with scalability and availability far superior to previous generations. Although IBM positions Shark as a monolithic system, it actually is monolithic in architecture only (i.e., “frame based”). Shark performs and scales adequately up to about 30TB, though most systems are in the 8TB-12TB range per frame. Although there is nothing to prevent an IT organization from linking numerous frames together, Shark often does not compete with the EMC DMX 2000/3000 nor the HDS Lightning 9980 V in very large-scale, transaction-intensive application environments. Nevertheless, Shark performs well for the vast majority of applications (90%+) and is worthy of shortlist consideration in those situations.
IBM also offers the Fibre Array Storage Technology (FastT) modular array systems, a product line OEMed from LSI Logic. IBM positions the FastT as a midrange system and the Shark as a high-end system, but in practice, the overlap is significant. This overlap is increased by the recent ESS 750 release. However, there are two key differences between the product lines. First, the 800-series is supported on both zSeries and iSeries processors, whereas the FastT is not. Second, the FastT has more limited software functionality, whereas the Shark supports the full range of IBM storage management software, including PPRC and FlashCopy. IBM offers cross-platform management of both systems through its SAN Volume Controller (SVC) and SAN File System (SFS) products.
SVC and SFS are IBM’s entry into storage virtualization. These products put IBM into the thick of this technology race, especially with EMC and HP (see Delta 2252). SVC supports some non-IBM equipment, specifically certain older HP and HDS subsystems. However, this support is presently more of a migration play than a heterogeneous management initiative.
IBM completely revamped its NAS line in 1Q04 by discontinuing its Windows-based NAS (which has not been particularly successful) and replacing it with a new NAS Gateway 500 product. This gateway is designed to provide both SAN and NAS partitions to either Shark or FastT systems. Although the NAS Gateway 500 product offers adequate functionality for most needs, it will most likely appeal to users desiring the convenience of an all-IBM solution.
Since the days when it considered NAS as the “be all, end all” solution to any storage problem, Network Appliance has done a remarkable job of evolving from a niche supplier into a Tier 1 storage provider. As shown in Figure 1 below, NetApp is the only company with positive momentum in every category in which it participates. Although it does not have a heterogeneous storage management platform nor an enterprise (monolithic) SAN solution with which to compete with EMC, HP, or IBM, NetApp is increasingly the “second supplier” in all of these situations. NetApp has proven to be unusually adept at assessing its own strengths and weakness and developing an appropriate strategy based on the realities of a given situation.
NetApp’s best-in-class NAS solution remains its flagship product and primary revenue producer. Within this product line, the company has built an impressive value chain based on software (e.g., Data ONTAP, WAFL file system, data replication). The company has also shown leadership in key areas, including backup/nearline appliances (e.g., NearStore), which currently is one of the hottest segments in the storage industry. The company has also been a leading proponent of iSCSI, offering free downloads of iSCSI drivers for its systems since March 2003. NetApp also took the unprecedented step of publishing all of its array management APIs so that its products can be integrated into other vendor’s management frameworks. This tactic is far preferable to the often meaningless API exchanges and reliance on laggard SMI standards.
As a pioneer of the ATA-based array market (NearStore), NetApp has also anticipated problems that are likely to occur as these systems scale with increasing large disk drives (e.g., 1TB by 2006/07). For example, the time to rebuild failed disks (extending potentially to days) combined with a large number of devices creates a statistical probability of a double device failure which will result in an unrecoverable system when using RAID 5. In response, the company has created a so-called RAID-DP, where the DP stands for “double parity.” Although this is similar to RAID 6 in concept, the company claims that overhead will be negligible (less than 3%) and will virtually eliminate the possibility of a double device failure without requiring the extra space of RAID 1.
NetApp’s major weakness is its inability to “play well with others.” NetApp is promoting its gFiler technology, which is capable of using a variety of back-end storage (presently HDS 9500/9900 V and IBM Shark). However, NetApp’s partnerships with other organizations rarely have been successful, largely due to channel conflict. gFiler shows great promise, but only if the company can solve its “go it alone” mentality.
Although StorageTek (STK) does not offer a broad enough product line to be considered an enterprise supplier (e.g., no NAS, minimal presence in the enterprise-class array market), its improved, targeted product focus has sharpened the company’s competitive position in key market segments. Specifically, STK has identified product areas that complement its dominant position in the tape market. Much of the development focus has been on storage application software (e.g., database recovery, content management), since the company has recognized the need to provide (and own) a value chain beyond the hardware platform. It is fair to say that the company has transformed a declining position in the SAN market into a much healthier position upon which the company can build and grow.
STK’s association with LSI Logic for the D-Series storage arrays has proved to be beneficial, giving the company a solid midrange product. The associated SANtricity software suite (e.g., snapshot, mirroring, remote replication) offers adequate functionality for most applications. Organizations most likely to consider D-series systems are those desiring a highly performing array in a standalone environment. The major drawback of the D-series in an enterprise setting is its lack of integration with other storage management systems (e.g., EMC, IBM, HP). However, the company has an OEM relationship with both FalconStor and Storability for heterogeneous storage management and storage resource management, respectively.
In addition to the D-series systems, STK has a strategic platform is its BladeStore product line. This product line is ATA/SATA-based, which is one of the fastest growing segments of the disk market. However, it also is a segment that is filling quickly with systems from almost every vendor. Therefore, complementary software will be critical to success. In this regard, STK is targeting data protection and archive applications (e.g., MirrorStore, E-mail Xcellerator, Application Storage Manager, continuous data protection and virtual tape). BladeStore is likely to be STK’s flagship disk product line, since it is the most complementary fit with STK’s tape-based data protection products. Moreover, these solutions are worthy of shortlist consideration for the areas they address.
The biggest question mark for STK is its mainframe product line, the V-Series. Although the SnapShot for V-Series offers some compelling space and total cost saving advantages, V-Series suffers from a lack of market presence outside of the most dedicated STK IT shops. Hopefully, the company will find a way to harness this unique snapshot technology in other areas of its product line.
Sun should be given credit: For the past 18 months its storage division has focused on addressing the needs of Sun-centric users rather than the quixotic notions of previous management to dominate the storage industry. After several years of incoherent acquisitions and dramatic annual strategy shifts, Sun management recently has focused on a steady course of bringing its acquisitions to market and shoring up its core product family. Indeed, the 6000 family of midrange storage systems offers significantly better reliability, scalability, and performance compared to the previous T3 family, and its OEM relationship with Hitachi Data Systems for the 9000 family gives Sun a competitive enterprise-class offering.
In addition to improving its hardware systems, Sun has finally brought its HighGround acquisition to full fruition as StorEdge SRM (Storage Resource Manager). More importantly, the company is close to realizing results from its acquisition of Pirus Networks. The Pirus acquisition, arguably its best storage acquisition, will enable Sun to manage not only its own disparate families of storage systems seamlessly, but also non-Sun systems as part of the N1 Data Platform. Sun’s strategic focus will be on midrange storage solutions and on reducing the cost and complexity of deploying storage. Sun has formed an alliance with AppIQ for some storage management functions, which creates a beneficial three-way relationship between Sun, HDS, and AppIQ. The N1 Data Platform will be the central component of its storage management integration strategy. Sun also positions its SAM-FS and QFS file systems as cornerstones of its ability to scale, provide high availability, and support clusters. However, Sun’s offering of three file systems (when including the predominant Veritas File System) sometimes results in muddled messaging and complicated systems management issues.
Sun’s current overall market positioning is as a “systems company” (a variant of the older “one throat to choke” mantra). Despite the company’s improvements in its storage strategy, Sun storage remains its weak link in the “systems” context, especially when compared to other systems vendors (e.g., IBM, HP). The 3000/6000 product lines do not offer sufficient technological advantages to justify a change from competing brands (e.g., EMC Clariion, HP EVA, IBM FastT, NetApp FAS 9xx, STK D-Series). Moreover, Sun has significant gaps in its product line, including ATA arrays, secure (compliance) storage, and NAS. Although Sun plans to address these gaps in the future, this situation demonstrates that the company continues to play catch-up and often lags its competitors by 6-12 months or more.
XIOtech has rightly enjoyed a reputation for delivering elegant and simple-to-use modular storage systems. More than almost any other storage vendor, XIOtech can claim to bring SANs to the “common man.” However, outside disruptions (e.g., a change of ownership from Seagate to venture capitalists, senior management changes, general layoffs) have hindered the perception of the company in the market place.
Notwithstanding the restructuring issues, XIOtech continues to focus on developing innovative technologies. The Magnitude 3D flagship product is the current generation of the Magnitude product line and is directed at delivery highly resilient, clustered storage systems. The effect of the clustering architecture is to offer loosely coupled “n-way” scalability over a wide area. The company claims significant productivity gains over conventional distributed storage methodologies.
As a midrange vendor, XIOtech does not offer storage management platforms or NAS products. It partners with complementary companies (e.g., Brocade, Oracle, CommVault), but does not offer interoperability with other storage management systems nor does it plan to do so. Nevertheless, for the specific applications that XIOtech addresses, its solutions are worthy of evaluation on a “best in class” basis.
Bottom Line: Since the release of the original SAN/NAS METAspectrum in February 2003, most technological advances have been incremental in nature. The "state of the art" has continued to mature, with surprising parity among the players.
Business Impact: IT organizations can select storage systems from robust competitive offerings, pursuing either a consolidated vendor approach to reduce storage management complexity or a "best in class" approach to optimize the efficiency of each technology area.
META Group originally published this article on 7 June 2004.