November 16th, 2004
HP's Fink finks on Solaris 10
In his blog that, judging by its first entry, is unquestionably designed to neutralize, and even perhaps counter the mud that Sun COO Jonathan Schwartz has been slinging at Hewlett-Packard, HP’s Linux vice-president Martin Fink has launched a frontal assault on Sun’s Solaris 10.
HP has been the victim of a series of vicious attacks by Sun’s Schwartz who, in his blog, routinely takes HP’s hardware and operating strategies out to the woodshed for a spanking. Eventually, HP had enough and sent Sun a cease and desist letter. But instead of ceasing and desisting, not only did Sun release HP’s letter for publication, it also publicly published a response that basically told HP to "go to Hell." Sun’s response also included the play-by-play on how Schwartz reached some of his conclusions. Since then, the lawyers from both companies either stopped talking to each other, or moved their communications to a private channel. My sense is, however, that HP’s threat was a bluff. The last thing it or any other company needs to do is launch a lawsuit that, by way of the discovery process, forces the the disclosure of confidential strategic plans.
So, what else was HP to do but launch a blog of its own? Enter Martin Fink — HP’s top-ranking Linux official. Strangely, Fink’s blog — called LinuxCIO.com — is not HP’s doing, but rather Fink’s alone. HP is clearly nervous about the blogosphere. The blog contains the standard "shoot-the-messenger" disclaimer that "The content on this site represents my personal views and opinions and is not necessarily representive of the opinions of my employer." Indeed, Fink will have to watch what he says. The last HP Linux executive that made a habit of expressing his own views in a public forum — Bruce Perens — ended up off the HP payroll. Though the split seemed amicable at the time, Perens told me over dinner in New York City that it "wasn’t his idea."
However, if Fink’s entry is an example of what we can expect from him, then the only people who will tune in will be the ones that are looking for the drama instead of a serious counterattack. Actually, tuning in will be difficult for that crowd, too. The site uses HTML frames (which went out of vogue last century) and has no RSS feed, as best as I can tell.
Perhaps it’s just a matter of some fine-tuning before Fink gets the mechanics right. But that’s precisely my criticism of what Fink has to say about Solaris 10. Most everything is a matter of fine tuning. Fink takes virtually every aspect of Solaris 10 to task to prove that it’s pretty much a non-starter. At the top of his list is a reminder that Solaris 10 for the x86 architecture doesn’t have nearly the library of applications behind it that Solaris for SPARC does, and that it’s not as though Schwartz can press some magical button to create the sort of momentum for Solaris 10 that Linux already has. While lack of applications is definitely an issue for any company attempting to jumpstart a new software ecosystem, it’s not like Sun hasn’t been honest about this challenge.
So, in essence, what Fink is really saying is that Sun simply isn’t up to the challenge, especially when you take into consideration the head start that Linux has when it comes to *nix on x86. On this point, I’m not so sure. For the applications that count — the enterprise ones like Oracle, BEA, and SAP that can put Solaris 10 in the game — those companies may have already embraced Linux. But to them, Sun has a proven track record for providing a gold standard *nix on which those vendors can depend when it comes to delivering solutions to their customers. And if Sun can deliver the reputation of SPARC Solaris (including performance, reliability, security, and features unique to Solaris) on the more cost competitive x86-platform at a price point that meets or beats what can be had from Red Hat or Novell, IT departments have no choice but to take a look. In terms of the thousands of other applications and open source projects out there that Fink cites as Linux’s difficult-to-overcome momentum, it’s not as though the source code for these applications isn’t there for recompilation. In fact, for many of them, that’s all there is. You have to compile for Linux, too. Trust me: If a particular application is causing some big customer to hesitate in favor of Linux, that application will show up on Solaris 10 on very short order.
While applications are Fink’s primary beef, the rest of his criticisms are small potatoes that question Sun’s sincerity when it comes to open sourcing Solaris and whether Sun’s Project Janus-based binary compatibility with Red Hat Linux will meet the needs of users who keep their Red Hat configurations up to date or want indemnification. On the open source front, I’m not so sure it matters. Sun need only deliver a package that’s technically better and more cost effective than its *nix competitors and paying customers will listen. Whether or not the full source for Solaris is available under the GPL or some other OSS license will not change Sun’s fortunes — or businesses’ interest — enough to matter.
In terms of Janus staying in synch with the Linux it must emulate, the same applications that will be the gating factor to Solaris 10’s success are also the ones that qualify to specific versions of Linux. The sort of update to Linux that would force Sun to update Janus in order to stay compatible is very likely the sort of update that those corporate applications wouldn’t be qualified to yet and thus, IT shops generally wait until their application providers guarantee support. Even then, a lot of regression testing is done before a new version is rolled out, which leads me to believe that time will not be an issue when it comes to making sure that Janus is up to speed.
As for indemnification, even Schwartz makes a big deal about it. But, as I said in a recent column, for end users, indemnification ain’t all it’s cracked up to be because suing someone that you’d eventually want as a customer doesn’t exactly make good business sense. Besides, Fink should be one to talk considering the limitations of HP’s indemnification program when it comes to Linux. (It only protects against a suit from SCO.)
In saying that "If Solaris is from Venus, and Linux is from Mars, sorry Jonathan, Venus is way too close to the SUN, I’m going in the opposite direction and to the red planet," Fink concludes that you’ll get burned if you get too close to Sun. However, after seeing Microsoft’s willingness to enter into a watershed deal with Sun, and the proclivity of forward thinkers like AMD and salesforce.com to become part of the Sun ecosystem, and a number of recent wins for StarOffice, the Java Desktop System and even Sun’s $1-per-processor per-hour-utility offering, I’m not hearing about any serious cases of Sunburn. I’m not saying that Sun’s strategy is a slam dunk either. But in the Schwartz vs. Fink war of the blogs, so far my scoring is 1-0 in favor of Schwartz.
