In-Depth

What Won’t Happen in Storage in 2006

10, 9, 8……let's see what the future holds.

• By Jon William Toigo
• 12/20/2005

I've been asked us to pull out my crystal ball and identify the Big Trends that will likely become the Top Stories of 2006. Always the rebel (and short of crystal balls), I refuse to engage in such a clichéd endeavor. Instead, this column focuses on the top ten things that I am reasonably certain will not happen in 2006. Proceeding à la Letterman:

Number 10:  Managed storage services and outsourcing will not become the dominant model for storage provisioning or management. Although companies have had it up to here with IT—and especially SAN—expense, you can’t outsource a problem. While some will try, the blush will be off the 10-year agreements within the first three years. A messy return to internal processing will ensue.

Number 9:  The mainframe will not go away, replaced by grids of cheap Linux servers. In fact, in many companies I have visited, there is renewed thinking about the efficacy of mainframe computing—if not as an application host, possibly as a well-managed front-end processor for massive storage in large companies. This model has wheels, because the mainframe world is the only place where you have virtualization, SRM, HSM, and true information lifecycle management today.

Number 8:   Tape will not die. Despite the onslaught from competitors in the disk world, tape still holds a place in low-cost, portable, long-term storage that simply can’t be beat. You will see an uptick in interest in compression technology (a sign of poorly managed data at the data level), in virtual tape technology (as a mechanism for off-loading backup streams more quickly), and possibly in encryption (to avoid another Bank of America incident), but all as a precursor to driving backup and archival data to tape. Tape is still cheaper than disk from a systemic point of view, though not always from a labor cost perspective. And portability rules.

Number 7:   Open-systems storage resources will continue to go largely unmanaged. No breakthroughs in SRM will occur until the vendors stop placing barriers in the way of management visibility, and that is just not going to happen. What you may see are some stealthy improvements in “SRA”—storage resource accounting. That’s my new acronym for 2006. Look to Zerowait, Monosphere, and maybe a few others to start providing tools that let you handle disk like inventory and model storage growth and allocation more efficiently. I hope to see Hi-Stor and ADSI do the same for tape inventory.

Number 6:   Disk drive breakthroughs are not expected. Perpendicular drives will ship next year, as will some interesting new drive technology from Seagate with a different form factor from what we see today, but with a massively enlarged capacity. Petabyte drives? Not now, but soon. The disk market never sees return on investment in drive improvements other than capacity. Seems that everyone believes that bigger is better.

Number 5:   SNIA’s SMI will not go unchallenged. The Storage Management Initiative found a quasi-partner/challenger with the announcement of an open-source initiative under IBM’s rubric this year: Aperi. However, it’s beginning to smell like Aperi may become too closely associated with SNIA to produce meaningful results (IBM is part of SNIA’s council, after all). If this happens, and even if it doesn’t, IBM has opened the door for other initiatives to do what SNIA—with its vested interest in the status quo—cannot. Question: can the open source community get all fired up about storage?

Number 4:   SAS/SATA II will not roll out with the same success as SATA I. There are a lot of things working against this set of standards. In addition to infighting over the efficacy of yet another set of standards that could effectively compete with Fibre Channel for corporate mindshare, the procedures for interoperability and conformance testing are showing signs of strain. Products showing up in the market from vendor X don’t seem to want to plug-and-play with products from competitors. That bodes ill for the adoption of this technology on as widespread a basis as we saw with SATA in 2004/2005.

Number 3:   Storage security will not become the hot market that the vendors and analysts would have us believe. Security is hot, owing to regulatory compliance matters; storage is lukewarm, but a necessary consequence of unmanaged data growth. However, the two are uneasy partners. Latency sensitivity is the hallmark of storage I/O, and security (encryption, especially) is still viewed as adding latency.

Even if it becomes more efficient to encrypt backups, restoring from an encrypted state will become another bugaboo. As for other aspects of security and storage, nobody seems to be paying a lot of attention to the vulnerability of the storage infrastructure from the standpoint of destruction or denial of service. Lastly, the new privacy laws appear to be getting watered down. Big companies are lobbying ferociously against the emergence of “draconian” laws and penalties for failing to protect consumer information from loss. They assert that part of the burden of responsibility should go to the consumer who placed his information in the company’s online repository to begin with.

Number 2:   Hardware-based content-addressable storage will not disappear. As much as we have railed against it and encouraged readers to wait until a software-based hardware agnostic solution appears in the market (which should come early in 2006), EMC and others will continue to leverage their one-stop-shop, one-throat-to-choke model successfully to sell “compliance in a box” to Flashing 12s in the front offices of financial and health care companies.

Number 1:   Storage will not get sexier. There is a well-documented drop-off in enrollments in university-level information technology programs in general. If there were ways to break these numbers down even further, I suspect that you would find a significant drop in the numbers of students who have chosen storage technology in particular as their field of interest. Disk is still disk: despite improvements in read/write heads, PRML algorithms, media coatings, or actuators. It is still about placing magnetic states onto rusted platters in a reliable way. In short, if applications and silicon are the Ferrari’s of tech, storage is the dump truck. If you were younger, which one would you rather drive?

Sexy or not, storage is essential, and in the coming year, given rates of unchecked, unmanaged data growth, it will only increase in importance and costliness to organizations. So, we will continue to cover it in this column.

Thanks to all who have offered feedback in 2005. Let me know what you are particularly interested in knowing more about in 2006 and I will do my best to provide it. Write to: [email protected]