- By Steven J. Vaughan-Nichols -
Does Sun Microsystems' CEO Scott McNealy dressing up like Tux and
proclaiming that "We love Linux" to financial analysts at Sun's
annual meeting on February 7 really signal a change in heart
by the Solaris giant?
If it does, it will signal an amazing turn-around. Back in the
May 12, 1999, issue of Computer Reseller News, McNealy
said, Linux is "a great way to get the wrong answer," even as he
announced that Solaris would run Linux programs.
Now, Sun is seeking the right Linux answer with a four fold
plan: 1. Create and ship its own line of Sun house brand Linux
servers; 2. Expand the Sun Cobalt Linux application server line;
3. Open-Source some Solaris' key components and release Solaris/Linux compatibility development tools and 4. Bring Sun's Open Net Environment (ONE) and the rest of its
Web Services initiatives to Linux. Indeed, with this move,
iPlanet Directory and Web servers, Java/XML, Forte for Java
development tools, Project JXTA, StarOffice, Chili!Soft, ASP
and Sun Grid Engine should all be moving to Linux.
That sounds great, but as Alan Gillen, IDC's research director
for systems software, comments, "It was a peculiar
announcement for Sun since it didn't have a lot of specifics and
it wasn't accompanied by hardware announcements."
Even without details, some companies that use Sun technologies
in their work rejoiced to see Sun moving into Linux. William D.
Vasu, president of CyberAccess, an information portal
company, wasn't "surprised at Sun's recent Linux moves. They
are a hardware company with a software heart and this move
arguably gives the customer a choice of hearts that will play
across their entire product line. Furthermore, it supports Java
and the open-systems services environment to the detriment of
the .NET onslaught from Redmond."
Why did Sun make this about a face? After all, as Gillen points
out, Sun has long dismissed IBM and HP with their multiple
hardware platforms and operating systems as serving only to
One reason is simply that customers want and demand Linux
now on the low-end. Even current Solaris users are finding at least "seven reasons" to consider Linux.
Stacey Quandt, Giga Information Group's Open Source analyst
thinks that Sun's making the move because in the short term, "it
will plug a hole in its product line by making Linux available on
two-way x86-based general-purpose systems." Specifically, it
will give Sun an alternative to Dell, HP, Compaq and IBM's
Intel-based general-purpose servers. For the telecommunications
and embedded business, the move will enable Sun to "offer a
Linux on SPARC alternative to Linux on Intel."
In the long run, Quandt goes on, "What is significant about
Sun's broadened support for Linux is that it targets x86-based
general-purpose servers and SPARC systems and ends Sun's
'last man standing' status in regard to only following a
proprietary Unix RISC road map strategy. Sun's decision to
augment its Linux strategy through organic growth with Linux
for low-end single-purpose appliances and low-end general-
purpose servers and Solaris on SPARC for high-value, general-
purpose servers, allows it to provide users with multiple Linux
solutions while maintaining the distinct brand value of Solaris
on SPARC for high-end servers."
HP's Linux business strategist, Mike Balma, thinks that this
might blow up on Sun. "It's nice to see them acknowledging
Linux can solve real problems, but doesn't this move put Solaris
on SPARC into question?"
Scott Handy, IBM's director of Linux software solutions, thinks
that Sun's Linux move has come too late. "We've been in Linux
for a long time. We'd know others would follow since Linux
has kept growing. Customers demand Linux and need it."
Balma adds: "Sun will be welcome to the Linux environment, they're coming
in very late. We've been supporting Linux for
three years, maybe in three years Sun will catch up."
Quandt, though, believes "Sun will do very well at Linux. Sun
doesn't want to create confusion for customers while it has been
a challenge for IBM to explain to customers when to deploy
Linux and when to deploy a higher end operating system like
AIX or OS/400." Still, Sun "could have done it sooner." Indeed,
everyone we spoke to was agreed on one point: Sun had to
move to Linux on the low-end. Business customers demand
Linux for file/server and edge servers such as Web and email
Officially, Sun has given out almost no specifics on its Linux
plans. Nevertheless, Sun insiders and analysts believe that some
of the foundation has already been laid. Here's what they predict:
First, Sun will go into the low-end server hardware business
with Intel-based boxes. It will not turn into a high-end imitative.
For example, Sun won't be producing four-way symmetric
multi-processing (SMP) or above servers on Intel. If you need
enterprise level computers, Sun will still be selling you Solaris
As further proof of this, Sun will not be delivering Linux virtual
machines (VM) the way IBM has with its iSeries, pSeries and
zSeries. Even as other Linux and hardware companies are betting that Linux VM will prove popular with business
customers, Sun continues to bet on Solaris on SPARC for this
The new low-end servers will be more than just Cobalt style
appliances. These will be turnkey systems running Sun Linux
and be sold through Sun's current partners. Sun Linux will be
very tightly coupled to these boxes' hardware.
While it's possible these will be white-boxes -- unbranded
systems built by other vendors -- it's expected that the Cobalt
Server Appliance business unit, under the direction of Vivek
Mehra, v.p. and general manager, will be expanded to meet the
expected demand for these boxes and a new series of Cobalt
appliances. We will not, however, see Linux workstations out of
Sun. Sun's Linux move is a server move, not a desktop one.
This move to Linux hardware boxes is may remind you of VA Linux
and Penguin Computing's hardware initiatives. Unlike these
smaller companies, though, Sun has large financial resources, a
world-recognized brand and a strong hardware reseller channel.
While this bodes well for Sun's Linux plans, if Sun is successful
it may put even more on pressure on Penguin and the other
remaining Linux hardware OEMs.
It's also expected that Sun "go-it-alone" Linux distribution path
may lead other companies to question their reliance on
partnerships with Linux distributors like Red Hat, SuSE and
That said, Sun isn't going it alone on Linux on SPARC. For
native Linux on SPARC look for new partnerships with Lineo
and SuSE. On the embedded side, look to Lineo and TimeSys, a
leader in embedded Java, for partnership announcements.
We also know for a fact that Sun will be producing a Linux Standard Base 1.1-compliant system as LSB continues to mature. Sun also plans to Open-Source some of Solaris' features and make them available for Linux. There is no word,
however, on what features might be Open-Sourced. It appears
however that Solstice DiskSuite, Solaris' volume manager, is
under consideration. In any case, no one knows yet under what
license Sun will release any currently Solaris-specific features.
Finally, despite any thing Sun might say about it possibly having
life, Solaris on Intel is dead. Solaris on Intel, which was often described by analysts and resellers alike as an operating system that only
existed as a bait and switch product for Solaris on SPARC, will
never see another major revision. You can expect to see current
Solaris on Intel users encouraged to move to Sun Linux on Intel.
Whatever the fate of Sun's Linux move; there can be no doubt
of one point. The move is of vital importance to Sun. As Quandt
says, "The absence of a low-end, general-purpose Linux strategy,
coupled with the erosion of Solaris on SPARC for edge
services, means Sun was faced with the prospect of defining
itself as a high-end performance company that specializes in big
iron. Sun's decision to confront market realities and discontinue
its focus on a single chip, single operating system environment
means that it stands a chance of reinventing itself."
question is, can Sun succeed in this reinvention?