If Cisco Systems announces its first blade servers on Monday, as expected, the news may well herald a major expansion of the dominant networking company's business. But even though it's the most hotly anticipated move in a long time for an IT vendor, this isn't the first case of a company taking a big gamble on entering into a new business.
Potentially game-changing shifts have taken many forms, and none is directly comparable to Cisco's plan or its historical context. But there are some lessons for Cisco in how those strategies have played out, according to industry analysts.
It may be hard to remember, but Intel didn't make any chips for servers until the Pentium Pro was unveiled in 1995. The company had remained focused on PCs while giants such as IBM and Sun Microsystems built both the central computers that ran enterprise applications and the processors at the heart of those systems. PCs were used for some departmental functions such as printing, but they weren't true servers. Leveraging its PC chip development resources and large-scale PC economics for server CPUs turned out to be a very good move for Intel and IT as a whole, spawning the industry-standard servers that dominate data centers today.
The Pentium Pro was a momentous introduction but wasn't as daring a move as Cisco would be attempting by entering the server market, noted Insight64 analyst Nathan Brookwood. CPUs were already Intel's core business. It was just a matter of bumping them up into a new class.
But another move by Intel didn't turn out so well. In the late 1990s and earlier in this decade, the company aggressively pursued communications, introducing new types of processors such as Arm-based StrongARM chips for handheld devices and even producing optical components for high-speed networks. Both efforts took Intel outside its core business and didn't gain much traction, eventually being sold off.
"I'd have to give them a failing grade on that," Brookwood said.
Some vendors can use their sheer size to make a daring foray succeed. Microsoft does this regularly, said analyst Roger Kay of Endpoint Technology Associates. The quintessentially dull software company tried to break into gaming hardware in 2001 with the XBox, and it struggled for years against established players Sony and Nintendo. Today the XBox 360 has a wide lead over its most direct competitor, Sony's PS3, but only after years of investment and losses. Kay doubts Microsoft has broken even yet on its cumulative investment in the platform.
"It's something Microsoft can do better than others, because they have the money and smarts," Kay said.
Success in a new business does require knowing what you're getting into. This is why Dell stumbled badly after it entered the printer business a few years ago, according to Kay. At the height of its success selling business and consumer PCs online, the company decided it could give Hewlett-Packard a run for its money in another mass-produced piece of hardware. What it didn't count on was HP's printing brand and intellectual property, as well as its marketing and distribution system, Kay said.
"They took a superficial similarity and assumed it was the same type of market, and it really wasn't," Kay said.
The biggest challenge generally isn't a new technology itself but the vendors that are selling it -- something Cisco would find no shortage of in the server industry.
"You may be the biggest banana in your market segment, but when you try and expand into another market segment, especially one that might not be tightly coupled with yours ... there's always going to be a big banana there already, and it's hard to displace entrenched, competent competitors," Brookwood said.
If it does enter the server market, Cisco will find a much stronger set of opponents than it has faced in enterprise networking. Though they all had technical strengths, the companies that once challenged Cisco head-on -- 3Com, Cabletron and Bay Networks (later part of Nortel) -- frequently miscalculated the market. The likes of HP and IBM tend not to.
Given that key lesson, analyst Tom Nolle of CIMI doubts Cisco will try building servers as they are traditionally defined.
"Cisco would be incredibly naïve to launch a general-purpose blade server in competition with IBM, HP and Dell, and I think their sales force would tell them that," Nolle said.
Because computing gear takes up a much bigger part of a typical enterprise's budget than networking, it would be an uphill battle for Cisco to dictate that piece based on the network, he said. "You do not want to fight against an IBM or a Microsoft ... in a big IT buy," Nolle said.
It's more likely the company would sell blades that aren't really servers but computing platforms that do specific jobs relevant to networking. This would be partly a defensive move. If Cisco doesn't bring those functions, such as collaboration and unified communications, into its own platforms, computing companies will have an opportunity to bring those in through their own server sales, he said.
Historically, the golden opportunities have been the ones where a vendor could break into uncharted territory, such as when IBM pioneered LANs with SNA (System Network Architecture) in the early 1970s, Nolle said. Big Blue then dominated that field into the 1980s.
"You can't find one where someone's jumped in where there was an incumbency," Nolle said.