Turbolinux’s turnaround game plan: Closed-source products included

Last week I said that Turbolinux was one of the handful of Linux pioneers who had begun to turn itself around in the last year, in the face of broad skepticism about Open Source as a business model. What has been the secret? I posed that question to a tag team
of senior managers at Turbolinux late Monday. According to Turbo’s
CEO, Ly-Huong Pham, they’ve begun to act less like a charter member of
the Open Source movement and more like a business — a traditional
software business. That business may be Linux-based, but that doesn’t
necessarily mean that Turbo’s going to be bound by Open Source rules or adhere to
an Open Source development model. That may be an ideological problem for Open Source enthusiasts. However, according to engineering v.p. and “extreme Linux” guru Peter Beckman, it seems to result in pretty good software. Having given Turbolinux’s new PowerCockpit software a test drive,
I think they’re right. Does quality trump ideology? Will Open Source
advocates allow closed source software to become an important part of a
Linux-based enterprise solution?

I began my Monday evening speaking by speaking to both Pham and her
newly announced marketing v.p., Dino Brusco. I asked them about the
perceived churn at the company, its current strategic direction and how they
planned to stay in business.

What was the company’s strategy for survival?

Pham: “When I was hired by the board 14 months ago, I did a
technical and strategic assessment of where we were. Based on what I saw and
where we were at, I decided that the best strategy was to become more
disciplined, operationally. I also recommended that we take Turbo from a strategy of
being a distributor to being a more rounded software company.”

According to Pham, that meant two things, being “more disciplined”
about the company’s distribution business and focusing more on proprietary
software development.

Disciplined isn’t a euphemism for walking away from the company’s
distros. It does mean trying to make that business rational and recognizing its
limitations. Pham says that “Turbo
6.5 had multiple distributors … distributions. It took a lot more
resources, in part because we didn’t have a disciplined distribution model, and we
didn’t deploy an [operating system] development model.” Pham also looked at what she
described as an “infrastructure issue.” “How could we support a global distribution infrastructure? We could by being far more efficient. We are, and we continue to be focused on operational discipline.”

Pham says this has helped the company “stay number one” in Japan,
its initial beachhead market. She says the company has gone on to make its
distro the top Linux platform in China. In addition to the Asia Pacific
region, it has come into the U.S. market, and earned a “top four”
position, despite a relatively late start here. And, Pham says the company has
done so without taking a financial bath. “I want to set the record straight.
The top four Linux distros have been under constant speculation about
their going out of business. Some of this comes from industry analysts, who
looked at public Linux companies, and they looked negative from a financial
perspective. We have been part of a larger industry that grew too fast, then faced a
downturn.”

She went on to say, “A lot of investments made by other companies
were nothing but attempts to buy market share. There was no follow up
[applications] investment — most assumed the revenue was going to
come from service.” Pham had trouble seeing how that revenue stream was
going to be sufficient. “Our model is going to be a software company.” She
says that the company had and would provide services, “but service is
not our core focus.”

So the company attempted to guard its funds. “We
are in a very financially viable situation. I wouldn’t be able to make
the investments I have in software development if that wasn’t true.
Unlike many other Linux companies, I’m not worried about having to go out and
raise cash.” Pham says that the company’s “discipline” meant that it
had the money it needed to bootstrap the development of enhancements and
enterprise-focused applications. Many of those products would leverage
Linux — and would be closed source.

Pham says, “I’ve done a lot of [operating system] business. I spent time at Apple.
There’s no big secret about how you develop an OS business. You invest at the
core level. You extend — you move to the middle tier. Look at our EnFuzion
product as an example. Then you move out into applications. ”

If that sounds like the business plan of another software company
based in Washington state, Pham is relatively unapologetic. She wouldn’t
mention the “M” word, but she did suggest that most Linux pioneers had been
“overzealous in promoting Open Source as a business model.” According to Pham,
Turbolinux “has always been a hybrid” mix of open and proprietary. She drew a
distinction between the open “development model” at the core of the OS and a more
closely held “business model … at the middle tier and up.”

This means that new products, such as PowerCockpit, will be closed
source. Pham justifies this based on feedback from the company’s customers,
especially those from China and Japan. “In some countries, businesses are more interested in
control of the solution, in having influence in the design and
delivery” rather than access to source code. That doesn’t mean that Turbolinux
won’t provide source to customers who need it and, presumably, would be
interested in paying for it. She says that issues about source would be “customer
driven,”but that the company would prefer to offer customer’s
access through an API — “a framework, where [customer-side] innovation is
more manageable” in terms of vendor support.

In addition, Pham committed the company to a multi-platform future.
She declared that the company wouldn’t limit itself to developing
exclusively on a Linux platform. PowerCockpit “will support different OSes. The
first is Linux, but other operating platforms will be supported including
Windows, and Unix, soon,” she says.

Pham also made a second commitment — to Linux and an open “core OS.”
She says, “We will continue to support Linux development. We will
continue to contribute to the community. We will continue to help develop the
core. It will be a two-way street.”

Linux advocates may object to closed source products and the core of
Turbolinux’s evolving strategy. Pham and Brusco are
counting on their new stream of products and partners to overcome that
resistance.

The centerpiece of those new products is PowerCockpit, a potentially important enterprise systems management tool developed
under the tutelage of “extreme Linux” guru Beckman, who was
recently made Turbolinux engineering v.p.

According to Beckman, PowerCockpit is an attempt to simplify deployment and support on a Linux platform in a way that should drastically reduce the cost of enterprise
administration. It also could cut costs dramatically for high volume distributors and
system integrators.

To oversimplify, Beckman proposes to use elements of Unix
architecture “that have been around for 20 years” to support the remote deployment
and upgrade of systems in a few easy steps. What Beckman is proposing is
not just remote installation of an OS or a piece of software, but an entire
“solutions stack.” This means that an enterprise manager could configure hundreds of boxes on a remote server farm, creating “proxy
servers in Beijing, firewalls in Tokyo, whatever you’ve got.” In some ways this
explains Turbolinux’s reluctance to embrace service as a revenue
stream. What they intend to do instead is automate and “commoditize” a big
piece of what had been a labor intensive customization and integration
process.

Beckman thinks that PowerCockpit can be a key to mass producing
“config-to-order systems.” As he puts it, he hopes developers will be able to
“push a button and generate one, or a hundred solutions boxes. One of
Beckman’s many objections to traditional Windows deployment is the
difficulty associated with deploying configurations across an enterprise. Windows
2000, for example, effectively requires new server boxes in order to
generate a clean implementation of the platform. Or it requires a lot of on-site
work. Neither are acceptable. By comparison, PowerCockpit is designed
to collect the “entire software stack” and reconfigure systems as needed.

It’s a great concept. What’s interesting is that it’s a
closed system, and that includes its design process. PowerCockpit wasn’t
designed in an open environment. According to Beckman, the company wanted to
avoid peddling futureware, even vaporware. It wanted a development program
that looked like traditional software development. PowerCockpit was
conceived, designed and developed with a lot of input from prospective customers,
but virtually none from “the community.” The result is a robust, proprietary,
closed source, possibly essential software product.

Just like that developed by a traditional software company.