Gartner logo

Enterprise Virtual Tape Subsystem 1H03 Magic Quadrant
16 April 2003
Nick Allen, Josh Krischer, Fara Yale

Document Type:  Research Note
Note Number:  M-19-6333
  Download PDF

The market is dominated by IBM and StorageTek, which are almost tied in terms of shipments during the past five years. New players, however, are providing enterprises with more choices.

What You Need to Know

Virtual tape is a maturing technology. However, although the products can be useful in data center tape consolidation, penetration outside the mainframe space is negligible. Enterprises should look at virtual tape as tool to improve performance and to reduce the number of physical tapes. Users — particularly European users — now have more choices with the addition of a new player, Fujitsu Siemens Computers, with interesting technologies, more scalability and more platform support.


The virtual tape market has evolved considerably since IBM first entered the market in 1997, but much less than expected. Almost all of the virtual tape activity has been in the IBM S/390 and zSeries mainframe market only, and there has been negligible activity in Unix and "Wintel" environments. We consider enterprise virtual tape vendors to provide support for IBM mainframes, other mainframes, Unix systems and "Wintel" environments. However, few vendors support all these environments. For this Magic Quadrant, we have only considered vendors that offer OS/390 or z/OS mainframe connectivity.

IBM and Storage Technology (StorageTek) continue to dominate this market, and our estimates of their shipments are shown in Table 1.

Table 1

IBM and StorageTek Virtual Tape Subsystem Unit Shipments

  1997 1998 1999 2000 2001 2002 Total Total VTS PtP
                Total VSM, Cluster VTSs
IBM VTS 382 573 510 502 535 310 2,812 574
StorageTek VSM 0 114 465 499 610 526 2,214 24
PtP = Peer-to-peer
VSM = Virtual Storage Manager
VTS = Virtual tape subsystem

Source: Gartner Dataquest (March 2003)

A new entrant in enterprise virtual tape is Fujitsu Siemens Computers (FSC). In addition, following its acquisition of Sterling Software, Computer Associates International (CA) now provides a software-only solution. EMC, Diligent Technologies, Bus-Tech and Luminex Software offer products that support moving mainframe tape data to disk via tape emulation, but do not migrate it to real tape. Therefore, we have excluded them from our analysis.

From the beginning, we predicted that this would be a market characterized by much "leapfrogging" in performance, and this has proved to be the case. Fortunately, most systems in the market now meet or exceed customer performance requirements. Our criteria and weighting for the Magic Quadrant (see Figure 1) are:

Vision Criteria and Weighting

Platform Support: 15 percent

Feature Attractiveness: 15 percent

Availability: 20 percent

Time to Market: 10 percent

Product Performance: 30 percent

Scalability: 10 percent

Execution Criteria and Weighting

Market Share: 30 percent

Marketing Effectiveness: 10 percent

Product Execution: 10 percent

Market Credibility: 20 percent

Financial Stability: 10 percent

Service: 20 percent

Figure 1

Enterprise Virtual Tape Subsystem Magic Quadrant

Figure 1

Source: Gartner Research (April 2003)

IBM: IBM was the first to market with a virtual tape subsystem (VTS) product and, over time, has done a good job of improving performance and function while protecting customer investments. With the addition of Fibre Channel Connectivity (FICON) support, IBM now holds the top position in S/390 and z/OS mainframe performance. It also added peer-to-peer configurations, which eliminate a single VTS subsystem as a single point of failure and also address disaster recovery issues, including the addition of a synchronous (immediate) copy feature to a remote peer. The acceptance of these configurations, despite their high costs, has been reasonably robust — roughly 20 percent penetration. IBM's robot that supports its VTS is its major weak point — most customers that have experience with IBM and StorageTek mainframe robots report there is practically no comparison. In addition, IBM's features for managing the disk cache, although much improved, were late to market and are less extensive than StorageTek's. Also, a single IBM VTS still has several small single points of failure. IBM also offers open-system attachments to the VTS, but its success in Unix and Wintel environments is quite limited.

StorageTek: Second to market was StorageTek's Virtual Storage Manager (VSM). Although it came almost two years after IBM's product, quarterly shipments are accelerating well. VSM is based on StorageTek's Shared Virtual Array (SVA) disk product. This is both good and bad news. It's good, because the SVA is a well-proven enterprise-class product with no single points of failure and with reasonable performance. It's bad, in that the VSM developers must wait for the SVA developers to make performance improvements, so the VSM developers are not in charge of their destiny. Because of this, StorageTek is without native FICON support and now trails IBM in performance. Thus, we often see StorageTek bid two smaller VSMs against one IBM VTS. StorageTek claims that it will ship native FICON support in early 2004 and that multiple StorageTek VSMs constitute one logical system and provide added redundancy.

Rather than peer-to-peer, StorageTek employs a "master-slave," "cluster" approach to eliminate one VSM instance as a single point of failure. Acceptance here has been mediocre, partly due to cost, but also because StorageTek provides more options for having the primary tape copy be remote from the VSM as well as the option of securing an additional copy of the data in a second tape library that can be local or remote. StorageTek is somewhat stronger in cache buffer management than IBM, and also offers more flexibility in import/export functions. Unlike IBM's VTS, synchronous copies in VSM clusters are not supported, although the "slave" VSM can support additional primary work. Currently, VSM does not support Unix and Wintel environments, but StorageTek plans to do so.

Fujitsu Siemens Computers: FSC, mostly only a VTS player in Europe, was essentially last to enter the virtual tape market, but with promising technology. Its CentricStor Virtual Tape Appliance offers the broadest platform support, such as mainframes with MVS, z/OS, BS2000 (FSC's own proprietary operating system), and major Unix and Microsoft operating systems. It requires no server-dependent software, works with all of the major backup software, and uses the widest variety of heterogeneous tape drives and libraries. FSC offers an entry model and a high-end system. Although it lacks FICON support, the high-end CentricStor model is arguably the most powerful external VTS in the market. FSC plans to add support this year for FICON, Small Computer Systems Interface (SCSI) Over IP (iSCSI) and serial Advanced Technology Attachment (ATA) disk drives. CentricStor also supports tape duplexing to a second tape robot, which can be local or remote. With a European installed base of about 90 units, FSC is now starting to establish a presence in the United States and is working with Fujitsu Ltd. to extend the market for CentricStor. Clearly, FSC will have to execute well on this strategy if it is to move out of the Visionaries quadrant.

Computer Associates: With CA's acquisition of Sterling Software came its Vtape software-only product. The product basically disappeared for a year during the acquisition turmoil. In late 2001, it re-emerged as BrightStor CA-Vtape, but we have seen little activity in our client base. For a while, it was the only product that allowed users to duplex tape output and to eliminate the virtual tape server as a single point of failure, because Vtape runs on the mainframe and, provided the user has a second mainframe, Vtape can be deployed on multiple mainframes. However, IBM and StorageTek have neutralized that advantage. In addition, the cache management features of Vtape were quite weak, and its performance was disappointing for more than moderate workloads. CA claims to have fixed these issues and is promising even-better cache management, but, judging from the lack of sales activity, it will be a while before Vtape is anything but a niche product. Nonetheless, CA claims that, since its acquisition of Sterling Software, it has increased the customer base of BrightStor CA-Vtape dramatically and that sales have increased 300 percent. It now claims to have hundreds of licensed mainframe customers and to have customer references using BrightStor CA-Vtape in organizations of all sizes.

Neartek: Originally a French company, founded in 1994, Neartek acquired More Solutions, a storage integrator based in Lakeville, Massachusetts, in 2002 and is now headquartered in Westborough, Massachusetts. Neartek has also brought in significant additional capital. Like FSC, its Virtual Storage Engine 2 (VSE2) offers broad platform and real tape support and is the only virtual tape solution we know of that supports the AS/400. Neartek sells a software-only solution and relies either on end users or on its indirect distribution channel partners to provide standard Wintel platforms and storage devices. The installed base of roughly 50 systems is all in Europe, but Neartek has recently begun to ramp up North American sales and marketing. Technically, the product has come a long way since its first inception, with the lack of FICON support the one notable missing feature. Neartek has a reasonable road map, including FICON support by early 2004.

Based on its small installed base, and the relative immaturity of the "new" Neartek, Neartek is clearly a niche player today, but with much promise if it executes in sales and marketing.

Outlook: We expect IBM and StorageTek to continue to dominate the OS/390 and z/OS market and remain competitive in shipments, but, outside of this space, new players will be able to increase their market share only by finding strong original equipment manufacturer or reseller partners.

Gartner's Magic Quadrant Methodology

Magic Quadrants measure the relative strengths of vendors in specific markets. Effectively, a Magic Quadrant maps a company’s current capabilities against its future promises (see Figure 2). Companies are included in the Magic Quadrant based on one or more of the following criteria: the number of client-generated inquiries about a company, a company's market share, or its introduction of a disruptive technology that has the potential to create a rapid shift in market share.

Figure 2

Gartner's Magic Quadrant

Figure 2

Source: Gartner Research (April 2003)

Actual vendor scores are determined by how well a vendor does in each area and how important each factor is in the marketplace. Factors and weights are regularly reviewed for their continued relevance, their importance in the decision-making process, and their ability to explain a product's success in the marketplace.

Key Issue
How will storage systems evolve during the next five years?

Acronym Key

ATA Advanced Technology Attachment
CA Computer Associates International
FICON Fibre Channel Connectivity
FSC Fujitsu Siemens Computers
SCSI Small Computer Systems Interface
SVA Shared Virtual Array
VSE2 Virtual Storage Engine 2
VSM Virtual Storage Manager
VTS Virtual tape subsystem