iSCSI in Exile

One year ago, iSCSI was the toast of the storage networking chattering class it was the glamorous ingénue making her debut. Now, iSCSI has been banished, locked away in a festering oubliette, chalking the days on the wall while it waits to learn of its fate.

What happened? The major storage systems vendors are more afraid than ever to fully embrace iSCSI (SCSI over IP), the technology that is supposed to bring the performance of Fibre Channel SANs to cheaper and less complex IP networks.

All of them – Dell Computer Corp. (Nasdaq: DELL), EMC Corp. (NYSE: EMC), Hewlett-Packard Co. (NYSE: HPQ), Hitachi Data Systems (HDS), IBM Corp. (NYSE: IBM), and Network Appliance Inc. (Nasdaq: NTAP) – are treating iSCSI as if it were a malodorous vagrant.

Without the iSCSI storage systems (the targets) that would drive strong customer demand, other iSCSI vendors – the adapter and switch vendors – are mostly stymied. So at the moment, everyone is sitting around, waiting for the likes of HP or IBM to fall back in love with iSCSI and recall it from exile.

Which may not happen for another year, or longer. Granted, some of the big storage providers, including HP and NetApp, do have iSCSI-enabled storage devices on their product roadmaps, slated for delivery sometime in early 2003. But among the big storage players, none is vociferously or lustily promoting IP SANs. And then there's Sun Microsystems Inc. (Nasdaq: SUNW), which is actively dissing iSCSI. It says that iSCSI is, more or less, evil. For at least the next few years, Sun has completely ruled it out (see Sun Says iSCSI May Be a 'Mistake').So how did iSCSI go from being the hottest, most eligible technology on the market – a Colossal Opportunity akin to the transition from minicomputers to PCs! – to being the hot stove nobody wants to touch?

Blame the wobbly economy, for starters. Enterprises say they aren't much interested in investing in new technologies, so the vendors are in no rush to burn R&D and marketing dollars only to be first in line with an iSCSI storage device that won't sell.

"What would really help iSCSI is if Alan Greenspan were to drop the interest rate to 1 percent," jokes Jeffrey Schnabel, VP of marketing at StoneFly Networks Inc., an iSCSI gateway startup. "Half the problem is the economy."

Also holding up iSCSI is the fact that the Internet Engineering Task Force (IETF) hasn't ratified the final iSCSI 1.0 spec. That's expected to happen sometime in September, and that development alone should put some grease to the wheels. Related to this are concerns – still – that iSCSI interoperability and performance isn't up to scratch.

Another theory floated by some industry watchers is that the storage systems guys are fearful that any new iSCSI products they introduce will cannibalize sales of higher-margin systems.Earlier this year, for example, EMC introduced Centera, a storage array that uses industry-standard components like IDE drives, Intel processors, and 100-Mbit/s Ethernet. But instead of delivering it at an entry-level price point with, say, an iSCSI option, EMC is selling Centera as a high-end box for a relatively nichey application (fixed-content archiving) starting at $200,000. Conspiracy theorists suggest it's because EMC doesn't want to undercut sales of its midrange Clariion family (see EMC Has Eyes for Huge Archives).

But the fundamental thing holding iSCSI back is that the storage systems companies are a cautious bunch: They've seen IBM get arrows in its back and don't relish the idea of immediately following suit.

Next: The Big Blue Whipping Boy

The watershed event that seems to have stalled the iSCSI market was IBM's utter lack of success with the TotalStorage IP Storage 200i. It was the first commercial iSCSI-enabled storage device on the market – and apparently, IBM couldn't find a single paying customer for it. Every one of the storage systems vendors I've spoken to in recent weeks sees IBM's 200i as the canary in the coal mine – demonstrating that the market isn't ready for iSCSI (see IBM Ditches iSCSI Box and iSCSI's Down But Not Out).

IBM declined to talk to Byte and Switch about where it's heading with its iSCSI strategy. Instead, the company emailed us a statement. Here's an excerpt:

IBM's statement concludes: "Eventually, as iSCSI matures and becomes native to the infrastructure, we intend to support that as well." Hardly a compelling endorsement for customers evaluating a new technology. At this stage, IBM is retreating into a defensive strategy. And this is the prevailing industry groupthink, largely the result of IBM's unhappy experience.

Some believe that IBM's mothballing the 200i was actually a healthy sign, serving to deflate the hype and help iSCSI move on the next phase. "The whole thing at IBM was actually a very positive announcement for them," says Doug Ingraham, senior manager of product marketing in Cisco Systems Inc.'s (Nasdaq: CSCO) storage technology group. "In terms of the hype cycle, iSCSI is moving down the curve... so we have to be very careful about how things get perceived."

That may be. But HP, another one of iSCSI's biggest early proponents, has IBM's travails fresh in its mind. "As far as the Tier One companies, there is a hesitancy" about introducing iSCSI products, says Bill Mottram, director of the infrastructure business segment in HP's storage solutions group. "There are going to be some forks in the road, as we saw when IBM discontinued the 200i." (Well, technically, the 200i is still available; it's just that IBM isn't actively selling it.)

HP seems to be back at square one. Before it decided to swallow Compaq, HP had said it would deliver a whole range of iSCSI storage systems and other Ethernet storage products by the end of 2001. Those plans evaporated into thin air. Now HP is dipping its toe... ever so cautiously... back into the iSCSI waters with plans to resell Cisco's 5420 iSCSI router, which connects FC SANs to IP networks (see HP to OEM Cisco's iSCSI Router).

To buy itself time, EMC is voicing its support for IP SANs – but it isn't backing up the rhetoric with products, yet. In June, at a meeting with Wall Street analysts, EMC CEO Joe Tucci pledged that the company "will lead in block-over-IP. We will lead in the transition to these networks." (See EMC Flashes Wall Street.)But Tom Lassen, EMC's manager of networked storage marketing, won't say when EMC expects to deliver iSCSI-based products, though he says his best guess about when iSCSI will really start to take off is late 2003 or early 2004. "As time goes on, we'll see ebbs and flows... iSCSI will run into a roadblock, then it will find some success, before it becomes stable," Lassen says. "It's been our experience, and it's been successful, to let the market decide."

Now, is that a case of leading the transition to IP SANs, or following it?

The bottom line is that, right now, the storage systems vendors don't see a competitive penalty associated with not having an aggressive iSCSI strategy, and they don't see much upside in putting one into place.

Here, for example, is Dell's take on the situation: "Dell is certainly not going to do anything that's going to precede the stability of the iSCSI environment," says Steve Luning, Dell's director of storage architecture and technology. "We're in study mode." And here's Quantum Corp.'s (NYSE: DSS) Lynne Van Arsdale, strategic marketing manager in its storage systems group: "All of the new technology innovations coming to market have been slowed down quite a bit, and that includes iSCSI."

In the meantime, Adaptec Inc. (Nasdaq: ADPT) has gotten so impatient with the dawdling storage systems vendors that it's planning to ship its own iSCSI storage device (see Adaptec Sees iSCSI Delayed).Who will serve as the catalyst to pull the iSCSI market into the mainstream? For now, all the likely candidates are mainly trying to make sure they don't step forward too early. And so, until then, iSCSI remains chained up.

— Todd Spangler, US Editor, Byte and Switch
http://www.byteandswitch.com