A great thing about the storage industry is that innovation can spring from anywhere. The trouble with the industry is that innovation does spring from anywhere.
By Dan Tanner
Enterprise storage buyers want to make the best purchase, but must also be careful not to go too far out on a limb. “Nobody ever got fired for purchasing (insert a leading brand name),” is the conventional wisdom. But the best product doesn’t necessarily come from the most respected brand. In fact, it’s rarely the case.
This is not to say that the better-known vendors don’t have good products. They almost always do. But true innovation tends to come from start-ups. Just read Innovator’s Dilemma, by Clayton M. Christensen, for a practical example from the disk drive industry.
Tough row to hoe
The row that the innovative start-ups have to hoe is an exceedingly tough one. Their innovations usually address the storage problems that are causing pain only in large enterprise IT departments, and often in only the most critical applications-not exactly the market segment where sales are made quickly or easily. And it’s also a segment that’s ruled by risk adversity. Not only must the start-up innovator convince potential customers that the technology is risk-free, but also that the start-up will survive.
For start-ups, the sales cycle is long and hard. Many determine that they can’t afford to sustain a direct sales force and turn instead to the value-added reseller (VAR) channel.
However, VARs must also bring in revenue and their salespeople are further from the start-up’s technical training and product knowledge than the start-up’s own sales force.
Why would a VAR be any more ready than the start-up to sustain a sales force through the long uphill sales struggle, especially when the VAR can sell “low-hanging fruit?”
The VARs that actually add value are real gems to start-ups and buyers alike.
If the start-up’s technology is truly innovative and useful (and not too far ahead of its time) one of the following happens:
- The company succeeds on its own. This is rare. It takes a fortunate convergence of market factors, patient investors, and perhaps luck.
- The start-up pushes its solution into the OEM channel, where it becomes a best- of-breed product that is sold by a trusted brand entity. This can also take a long time, because trusted brand vendors are cautious.
- The company is acquired by a trusted brand vendor. Whether the product makes it into the market is now up to the trusted brand vendor, which could have bought the company to quash the technology in order to milk its own cash cow a bit longer.
- The start-up fails. In this case, very often the company’s intellectual property is sold off to a trusted brand vendor.
What’s interesting here is that in each of the four outcomes the technology survives. Let’s look at the record.
IP storage networking
Fibre Channel came about in the day of 10Base-T Ethernet as a gigabit-speed solution for the backup window. At the time, pundits said that IP could never be used for storage networking. Start-ups in the IP storage networking market have succeeded (very few), been acquired (a few more), or failed. Now, Fibre Channel at 2Gbps and 4Gbps is chasing IP at 1Gbps and 10Gbps. TCP/IP and iSCSI protocol offload engines and host bus adapters (HBAs) are widely available. On both a capital and operating cost basis, IP wins over Fibre Channel, and a comparison of the market sizes (Ethernet/IP and Fibre Channel) makes the eventual outcome a no-brainer.
Storage virtualization, originally viewed as a way to break large vendors’ strangleholds on users, was a great idea but was combated by those large vendors; adoption was minimal due in part to integration, support, and management issues. But a few of the start-up virtualization vendors have adapted by offering improved management tools and by broadening their solution portfolios.
Interestingly, file-level NAS and block-level SAN have complementary pluses and minuses, so a solution that converges NAS and SAN attributes (keeping the best and discarding the rest) is an industry “Holy Grail.” But any such solution (and similar ones, such as SAN file systems, wide-area file services, and universal file namespace solutions) can appear to users to be a big technical risk and/or require proprietary hardware. And if users are miffed about proprietary lock-ins from big vendors, how likely are they to accept the same from a start-up? Most of the NAS-SAN convergence players have failed, a few have been acquired, and some intellectual property has been snapped up by larger vendors. But products from the trusted brand companies have been slow in coming.
Tape backup is complicated, unreliable, and the (serially accessed) data is nearly always offline to applications and users-and in some cases can be unrecoverable. Nevertheless, tape is still the primary backup/recovery medium. But tape just doesn’t scale to meet today’s backup and, more important, recovery needs.
Tape’s role should be the media of choice for long-term inactive archiving, and disk-based backup should move to the fore. The reason why start-ups in this arena are having so much trouble gaining traction is a mystery. Perhaps there are too many technology choices, ranging from compatible virtual tape devices to proprietary disk-based appliances, which confuses end users.
Continuous data protection
Last year, analysts at a storage industry event pooh-poohed the idea of continuous data protection (CDP). This year, at the same event, the same analysts were calling CDP a panacea for everything. Nonsense. Users have to plan and classify their data and then decide on the basis of their recovery point objective (RPO) and recovery time objective (RTO) for that data whether to apply CDP. For all other data, traditional backup or even archiving is probably sufficient.
CDP as well as ILM will catch on. The difference is that CDP can be productized (now, typically for e-mail or databases), while ILM is a process that requires planning, data classification, storage tiers, and best-of-breed hardware and software selection. No single vendor can truly offer ILM, but many will market ILM “solutions”-and the trusted brand vendors may actually get away with that. CDP, however, is a product and it can be marketed. But the CDP space is already overcrowded, and there will surely be a shakeout.
In my opinion . . .
IP will become the storage networking lingua franca; all storage will be virtual; management tools will become standardized; backup will become disk-based and tape will become purely an archive media; storage will become a utility resource and files will be quickly and securely available from just about anywhere; data will be hosted where it belongs (ILM) and protected continuously (CDP). The step after CDP will be CDA-continuous data availability.
Dan Tanner is an independent industry analyst and consultant, and founder and principal of ProgresSmart (www.progressmart.com).