Enterprise Unix Roundup: Plateau or Bubble?
November 3, 2005
At its core, the model is based on the belief that all technologies go through five phases: technology trigger, peak of inflated expectations, trough of disillusionment, slope of enlightenment, and, ultimately, a plateau of productivity.
It's easy to see how the Web itself has followed this model. From the creation of MOSAIC to the .dot-com Boom and subsequent Bust, to the recent resurgence in venture capital interest to technology companies focused on product as opposed to concept.
On the other hand, the phrase, "Web 2.0" was frequently bandied about, and the Swag Scale, our gauge of booth handouts, was pretty high. It seemed the former 20-something-survivors of the dot.com revolution, were out in full force, now grayer and more sanguine as they attempt to drive the open source movement toward profitability. This observation was confirmed by MySQL CEO Marten Mickos, who noted, "survivors of the Bubble are successful and are running open source software."
However, not every undertaking is a star. Of the 100,000-plus projects on SourceForge, according to Conference Founder and Program Director Matt Asay, close to 85 percent are not actively developed; 80 percent of projects have fewer than 10 active leaders; and 73 percent have only one active developer. Of all these projects, an even smaller number are achieving commercial success.
A less-promising statistic, cited by David Skok, a general partner at Matrix Partners, is that the revenue from an open source software sale is typically 10 percent to 15 percent of a proprietary ISV sale. However, ISVs selling open source software generally have "a quicker ramp-up, lower costs, and it is evident earlier which will be successful."
We have no doubt that community has much to do with the first two advantages, and we can only hope the third prevents the irrational exuberance of a secondary bubble, where pet food by mail will change the world.
Skok further noted that many of the smaller percolating projects and companies will likely be consolidated into Red Hat, JBoss, MySQL, and SUSE. He said he believes the expansion within a contracting market may lead to the commoditization of open source software.
The open source movement has been transformed since it was renamed from the Free (as in freedom, not beer, as was oft repeated) software movement in the late '90s. These days, the more opportunistic members have swapped ponytails for suits, and the influx of ISVs and OEMS, not to mention venture capital firms and systems integrators, makes this consolidation and shifting seem perfectly natural.
The community concept is still a strong force, however, as sessions about upcoming creation of the GPL v.3.0 and discussions of product support revealed.
Ultimately, however, open source, Sun CTO Hal Stern said, "is about a mechanism, not a product." Stern said that "initial innovation primarily comes from places where it is paid for," (i.e., companies and educational institutions). Incremental innovation is also important, however, and a substantial portion of this innovation will come from places without remuneration.
Enterprise interest in open source software correlates directly with how well a given app meets the organization's needs. Its origins are but one fairly insignificant facet of it, as the members of one panel said, jokingly referring to their session as, "The Open Source Dating Game."
Three open source software firms and four financial services companies and one governmental organization faced off. The CIOs revealed their pain points, and the open source software firms pitched their products. The CIOs then responded with their concerns.
Their top concerns and interests had little to do with much of the noise surrounding the open source movement. They were not looking strictly at cost savings; nor were they particularly concerned about IP issues or inherent security risks.
There was near-unanimous sentiment that open source software is never free, and its deployment is not always less expensive. Those in the ISV community concur. On another panel we sat in on, Joyce Park, founder and CEO of Renkoo, noted that many enterprises will chose an open source stack not because it's free, but because it allows for customization. The customization brings with it fees, however, as the developers employed to do the customization must be paid for their work.
What the CIOs did worry about, however, were issues that are equally prevalent in proprietary applications: vendor and product longevity and what happens when either is discontinued, how much maintenance the product requires and how complex it is, how well a product integrates with the current infrastructure, and how mature the product is.
As we all know, Linux, although the underpinnings of many open source products, is no longer (if it ever was) synonymous with open source. From our vantage point, both at the show and at home, Linux is more often than not the operating system of choice for the ISVs writing apps and enterprises making a platform purchasing decision.
Microsoft brought a contrarian point of view to the table. Keynote speaker Jason Matusow, from Microsoft's Shared Source Initiative, said "As open source commercializes, it becomes less open. Commercial intent means vendors needs to lock things down."
"Open source," he continued "is equally proprietary. Someone still owns it and determines who uses it and how."
While the less sharp in the audience may have had their heads spun round by this argument, which seems sensible on the superficial level, our head stayed firmly in front.
We agree, though with Matusow's statement that it's not the release of code itself that is of value, but what the code does, "Released code has to be compelling. Throwing out garbage code does not go over well." Okay, that's a no-brainer, but he loses us with his assessment that, "Companies don't want to modify Microsoft source code, and they feel the same about Red Hat and SUSE."
Sure, companies, may not want to, but ISVs surely do.
And as Intel's Dirk Hohndel noted, "competition leads to innovation. Good-enough solutions from open source vendors with a different economic model has resulted in a proliferation of innovations from the proprietary vendors."
Our takeaway from the show: There's no denying open source software is going, if not has gone, mainstream. And it's not getting there strictly on its earlier strengths alone. The cost savings and community atmosphere may have gotten open source software onto the edge, but its customizablity and knowledge base are getting it into the data center, where its success will be dependent on the same factors as proprietary commercial products.
Savvy ISVs will recognize that open source is more a vehicle than a rally cry or a buzzword, and a solid road map is necessary for for success. Enterprises that have shied away from open source software up until now would be wise to begin test driving applications, at least on the network edge.
We're not sure we buy priceline.com CIO Ron Rose's prediction that "five years from now the application space will be all open source companies and managed service providers," but it is certainly something to chew on.
» Well, it's not like we didn't see it coming. Late Wednesday afternoon, Novell announced the pending layoff of 600 staff members worldwide. Novell declined to comment on this cutback right up until the announcement, which is pretty much its standard operating procedure. When the story broke, about the 200 layoffs in its less-than-optimal EMEA division earlier this year, the news had been leaked weeks ahead of time.
Leaks aside, Novell must get its act together, and soon. Beyond letting these people go, the company says it plans to rebuild the company around its Linux and identity businesses. Which leaves a lot of folks wondering just what will be shed to make this happen. Celerant, Novell's consulting subsidiary, is definitely on the way out, but the fate of a lot of projects is still up in the air.
» HP this week became the first vendor to unveil an Itanium 2 blade. The BL60p is its first Unix blade and runs only HP-UX. It can be housed in the same chassis as Opteron and Xeon blades, which will save real-estate and simplify server management. We're honestly not sure what to make of it. Our best guess is that it was originally a customized solution that now provides HP with another avenue to sell both Itanium 2 and HP-UX.
» With little fanfare (and who could blame the company?), SGI announced this week it is getting delisted from the New York Stock Exchange and will now be traded as a "penny stock."
Where this leaves the high-end server company is anyone's guess. SGI made some inroads into the cluster space a couple of years ago when it launched the Altrix server line, a set of Itanium-based servers that used Linux or IRIX to deliver high-end clusters to their customers. The problem is SGI has yet to try leaving the hyper-specialized verticals (i.e., engineering, technical, and computer-generated graphics) and sell its wares somewhere else. Hey, we know how it is to have a comfort zone, but with IBM and Sun eating up the high-end and midlevel cluster markets, now is the time to take a few chances.
» Red Hat, after a long bit of silence that we in the media tend to describe as normal for it, popped out of its shell this week to announce an inititive to get the Xen virtualization engine into the Linux kernel.
This has been done before, mind you, but earlier efforts had fallen by the wayside. By getting Xen integrated into the kernel, the virtualization technology will allow many more integration possibilities for Red Hat in its target enterprise market.
Elsewhere in the Corral
Recent relevant articles about enterprise Unix