Fool’s Gold

It’s human nature to get overexcited about “the next big thing,” and wireless is no different. With Intel about to debut Centrino, its new laptop Pentium that embeds WiFi directly into the chip, the tipping point is coming fast. According to some prognosticators, we could see over 40 million WiFi devices in less than five years.

Sound familiar? After Microsoft added TCP/IP to Windows 95, suddenly, anybody could get on the Internet — and did. With hindsight, we know that the sudden presence of tens of millions of new web users didn’t necessarily create new Internet-based industries, but prodded existing businesses to add new online channels. The biggest change was forcing companies to get serous about multichannel strategies, blending the customer experience, regardless of whether the interaction transpired via phone, web, email, chat, snail mail, or face-to-face. And, the economic downturn notwithstanding, that has created new opportunities for companies to add new value to existing customer relationships.

Ted Schell, chairman of Cometa Networks (a WiFi network wholesaler startup backed by IBM, AT&T, and Intel) predicted a similar future for WiFi, viewing it as a natural extension of the mobile workplace. He insisted that WiFi providers would succeed only if they built services that could fit into existing carrier offerings, such as VPN, rather than create the newest iteration of a CLEC (competitive local telecom exchange carrier).

According to Schell, WiFi providers targeting corporate markets must guarantee carrier- grade service levels and security, embed connectivity within existing corporate VPN offerings, and make the service available at distances no greater than a 5-minute walk or ride from where employees are likely to be during the workday.

As we noted a couple months back , WiFi will be the next little thing, with hotels a logical starting point for offering services. In this economy, it can become all too easy to hype any new technology that promises market sizes in the tens of millions. Yes, it might create a new opportunity for personal firewall vendors. However, more disruptive technologies like VoIP might be enabled by WiFi, but not exclusively tied to it. Like the Internet before it, WiFi won’t change everything, but instead, add valuable extensions to what’s already there.

Seeping From the Edge

Like most tech conferences, Linuxworld carried two levels of messages: (1) cool tools for system admins, who comprised the bulk of attendees, and (2) strategic messages of press releases and keynotes. In that sense, Linuxworld was no different than, say, early Java Ones — where crowds focused more on cute client-side applets instead of the appserver technologies, where the real action was emerging.

Consequently, our biggest impressions came, not from the show floor, but from pronouncements about corporate Linux strategies: the CIO from Unilever, announcing that Linux would become their next-generation enterprise platform, or Morgan Stanley, reporting over 30% of its core trading applications migrated to Linux. Typically, Linux supplanted UNIX; if we were paranoid, it would tempting to view Linux as simply another secret Microsoft conspiracy.

OK, we know that’s far-fetched, but we’re also not the only folks who view UNIX as the prime target for Linux — a widely cited Goldman Sachs study reached the same conclusion. However, the importance of these announcements wasn’t as much about UNIX as it was about history repeating itself. Remember companies –like Unilever — committing to UNIX at the dawn of the client/server a decade ago?

To date, Linux has primarily emerged on “edge” servers performing network-oriented web, file, and DNS serving. At this point, the question isn’t whether the big ISV household names will embrace Linux, but when. Red Hat noted that the biggest hurdle to Morgan Stanley’s Linux migration was getting the customer’s main ISVs — Oracle, Veritas, Rogue Wave and others — to synchronize on the same Linux kernel version. While the free wheeling nature of open source has gotten Linux this far, for enterprise Linux, platform players will end up setting the de facto standards.

Nonetheless, open source has opened a few cracks in the fa├žade, enough for JBoss and MySQL to gain critical mass with cheaper, quicker alternatives for J2EE appservers and databases.

Otherwise, for the platform folks, Linux has provided IBM, HP, and Dell golden opportunities, while thrusting challenges at Sun for keeping Solaris relevant. We stubbornly believe that Sun must fight fire with fire. Itanium’s slow pickup in the marketplace provides Sun the opening to make SPARC the definitive platform for 64-bit Linux, not to mention the strategic destination for the Solaris customer base.

Sunrise Sunset

As Year 2003 begins, we believe that this will be a do or die year for Sun.

Sun 2003 scarily resembles Digital 1993. Consider the similarities:
* Both competed on single, unified architectures offering cheaper, reliable mainframe alternatives.

* Both claimed huge ISV support and promoted best-of-breed solutions.

* Both pushed TCP/IP. Ironically, had Digital played its cards right, it should have become THE
Internet company.

* Both faced similar disruptive challenges that they initially dismissed: UNIX for Digital and Linux for Sun. (We believe that Linux, not Windows, is currently Sun’s biggest threat.)

More importantly, both companies had respectable, but neglected software portfolios. Digital, which practically invented groupware with messaging and collaboration products leveraging DECNet, but refusing to sell it on Windows platforms. Sun’s Java commanding the agenda for web development, but squandering its advantage by picking religious fights with Microsoft and IBM.

Sun never knew how to make money with Java, bungling the acquisitions of NetDynamics, Netscape, and Forte. With IBM’s Eclipse making a mockery of the Java Community Process, IBM has largely preempted Sun in driving Java de facto standards. Fortunately, with Jonathan Schwartz, Sun software finally has adult supervision — but has it come soon enough to avoid DEC’s fate?

As Sun admits, its real strength is systems. Although under siege by “Lintel” (Linux on Intel), SPARC still has breathing room. For now, so does Solaris. However, with IBM and the open source community pushing hard, Linux will likely close the 80% of the functionality gap within 18 – 24 months. Sun’s response — to embed Sun ONE into Solaris (alongside WebLogic — go figure?) is a year and a day late. While Sun’s N1 vision of autonomic data centers is based on future Sun ONE software, until this point, Sun ONE has been a costly distraction and drain on limited revenues.

To survive 2003, we propose a not-so-modest six-step program:
1. Dump Sun ONE. Strong medicine yes, but remember, most Java folks are buying
WebSphere, WebLogic, or open source alternatives.

2. Continue bundling BEA WebLogic lite (and throw in Borland JBuilder), elevating them to preferred status.

3. Open source Java to counterbalance IBM’s technology dominance over Eclipse. Do the same with N1 to cost-effectively tap the community’s latent R&D capacity.

4 Embrace the heck out of Linux, bulking it up with Solaris capabilities. Optimize Linux on SPARC.

5. Direct all resources into what Sun does best: Building industrial strength systems and doing best-of-breed optimization on Sun platforms. Do this in conjunction with the Accentures of the world, who are threatened by post-PwC IBM. Make that integration simpler, cheaper than IBM or Microsoft.

6. Finally, merge with Fujitsu Siemens, the “second source” of SPARC hardware, which has good EMEA presence and, more importantly, a fast-growing services business.

We’d like to be proven wrong, but to do so, NetBeans must draw real ISV support, Sun must stop contesting every IBM/Microsoft XML web services idea, and its internal organization must learn how to spell software. Otherwise, Sun’s best alternative is becoming a simpler, cheaper, less body shop-focused Linux alternative to IBM, because IT shops demand a second source.