Today, mainframes and tape continue to be closely linked in the minds of users and in data centers around the world. The mainframe has long used tape for backup and long-term storage, while Unix/Linux and Wintel platforms have recently become more partial to disk. Giant mainframe (IBM) and large-tape storage (IBM, Sun Microsystems/StorageTek) vendors continue to see healthy revenues and profits while moving the technology ahead steadily. That is why, of all the tape-plus-mainframe news passing across my screen lately, three items loomed largest:
- Oracle, which recently acquired Sun and, therefore, StorageTek, indicated that it will
- continue to invest in StorageTek's high-end/mainframe tape solutions, rolling out a new generation of products.
- IBM extended its Diligent deduplication technology to its mainframe TS 7680 virtual tape library (VTL) product.
- At EMC World, EMC announced its "Tape Sucks" campaign, seeking to persuade customers to deep-six tape and VTLs in favor of deduplicated disk.
What do these three items have in common?
In my mind, the common underlying message was the sense that many customers increasingly think of tape as a "second-class citizen," taking a back seat to disk and solid-state. Thus, Oracle's message could be seen as a reassurance to concerned tape customers that StorageTek will continue to advance its technology in the face of the Oracle-Sun acquisition, which seemed to promise ruthless elimination of old-style technologies that offer revenue declines in the next few years. IBM's extension of dedupe to VTL seems to suggest that it wishes to reassure customers that tape is still a fruitful market for Big Blue. Meanwhile, EMC's new marketing campaign suggests that, as a player firmly committed to disk instead of tape, EMC believes people are ready to switch and now is the time to make some serious inroads into the tape market.
The problem with all this anticipatory action is that, technology-wise, tape continues to advance quite nicely. The LTO roadmap provides clear evidence of this in the midrange, and IBM's TS 7680 offers up to one petabyte per solution -- 25 PB of equivalent "undeduped" storage -- hardly the sign of a product lacking scalability.
Moreover, deduplication technology does not necessarily change the equation. As a youngster, I read a rather odd set of legal short stories by Arthur Train, whose constant refrain was, "What is sauce for the goose must be sauce for the gander." Or, if the law applies to you, it applies to me. The deduplication technology "law" applies equally to tape and disk; dedupe-driven improvements in disk storage capacity and performance should be matched by approximately equal improvements in tape capacity and performance.
That leaves the ongoing capacity strengths of tape as well as one other factor: Tape remains far less energy-intensive than disk (despite recent improvements in the ability to "quiesce" disk while it is not being used). As storage needs increase by 50% per year and concerns about carbon emissions rise, the advantages of tape versus disk are actually increasing in some situations, such as 10-year archiving.
In other words, Oracle's reassurance to StorageTek customers should be seen not as a hedge-your-bets strategy in case old-fogy customers still want to cling to the merrily rotating cartridges of their youth, but as a canny realization that there is indefinite life ahead for tape -- and for the mainframes that use it most effectively. Tape does not suck; it is not even a cash cow. It is, simply, a storage tier that will continue to provide benefits that other media cannot match in situations that will continue to be important to the business. StorageTek is not going away, and neither is tape.
ABOUT THE AUTHOR: Wayne Kernochan is president of Infostructure Associates, an affiliate of Valley View Ventures. Infostructure Associates aims to provide thought leadership and sound advice to vendors and users of information technology. This document is the result of Infostructure Associates-sponsored research. Infostructure Associates believes that its findings are objective and represent the best analysis available at the time of publication.
What did you think of this feature? Write to SearchDataCenter.com's Matt Stansberry about
your data center concerns at firstname.lastname@example.org.
This was first published in May 2010