|
|
|
Market Roundup July 25, 2003 |
|
|
|
SCO Raises
Linux Stakes: Five-Card Draw or Three-Card Monte? Pressure Drop: Sun Axes LX50 Entry Level Server |
SCO Raises Linux Stakes: Five- Card Draw
or Three-Card Monte?
SCO announced this week that it has received U.S.
copyright registrations for UNIX System V source code, a jurisdictional prerequisite
for enforcement of copyrights. The company also announced that it will offer
UnixWare licenses tailored to support run-time binary use of Linux for all
commercial users of Linux based on kernel version 2.4.x and later. According to
the company’s claims, since 2001 commercial Linux customers have been purchasing
and receiving software that contains misappropriated UNIX System V source code
owned by SCO. In particular, the company pointed to capabilities in the Linux
2.4.x and 2.5.x development kernel that allow Linux to scale to thirty-two and sixty-four
processors through the addition of Symmetrical Multi-Processing (SMP)
capabilities taken from UNIX System V. According to SCO, the company will hold
harmless commercial Linux customers who purchase a UnixWare license against any
past copyright violations and of any future use of Linux in a run-only binary
format. In a related event, Open Source Victoria (OSV), an Australian Open
Source group, has asked the Australian Competition and Consumer Commission
(ACCC) to investigate SCO’s claims for
‘misrepresentation of need,’ where an organization suggests that people must
make payments that, in fact, they are not obligated to. The OSV likened SCO’s offering of UnixWare licenses to a “Nigerian scam or
Internet extortion ploy.”
Just when you think SCO’s
meandering effort against IBM could not get any weirder, the company delivers
yet another claim that reads as if it were slapped together during a
particularly liquid happy hour. The most curious thing about this latest
directive is that it reads as if the company’s lawsuit against IBM had been
decided in SCO’s favor. In reality, the case has not
even been accorded a preliminary hearing let alone a trial, meaning that
evidence of SCO’s claims has not been seen by IBM, a
judge, a jury or the general public. Since SCO is demanding license fees for
what are for now essentially fictional UNIX copyright infractions, one must
wonder if the company will refund whatever license fees it collects should it
lose its suit. One seriously doubts it. More likely, SCO sees collection of
these license fees as proof of its claims that it may need to use to bolster
its position in court. Another odd bit concerning this press release regards SCO’s comments about the SMP capabilities of the Linux
kernel 2.4x and 2.5.x. Since the vast majority of
commercial users have deployed Linux on lower-end servers, where SMP is
superfluous, why would they need SCO’s UnixWare
licenses?
Maybe they do things differently in Linden, Utah, which may be closer to the Old West than is good for SCO. From its public actions, SCO seems to see itself as a tough hombre in the Clint Eastwood mold, outnumbered by powerful adversaries but willing to risk all for the sake of intellectual property. But from the outside, SCO appears to be little more than a goofy adolescent desperate for attention. The company’s previous attempts to spread FUD among IBM, AIX customers and the Linux community have elicited little more than a jaunty “See you in court,” an inevitability SCO seems anxious to avoid and from our point of view, for good reason. From where we stand, the OSV’s request for a government investigation into SCO’s UnixWare licensing scheme seems a reasonable and appropriate response. We only hope that members of other Open Source communities are willing to take similar action. Enough already.
IBM announced this week that it is including grid
computing capabilities in the latest version of WebSphere Application Server
V5.02 - Enterprise. The new software comes from IBM’s recent acquisition of
ThinkDynamics, a company that specialized in the management and load balancing
of clustered servers. The new WebSphere feature will allow IT managers to
install what IBM calls a “traffic cop-like” component that automatically
monitors application workloads and then routes traffic to different servers as
needed. The company said the latest version of WebSphere now supports twenty-six
separate platforms, including the recently added Windows 2003 Server. IBM said
WebSphere Application Server V5.02–Enterprise, including the new grid computing
tools, would be available this week for a price of $30,000 per processor.
Grid computing is a relatively easy concept to grasp. If
one considers the desktop SETI project — wherein thousands of volunteers
allowed their desktop computers to be used to sort the millions of radio
signals emitted by the universe and collected with radio telescopes — one has a
pretty good idea of how unused computing horsepower can be utilized across a
wide area network. Grid computing as envisioned by IBM offers a more mundane
and practical application of this capability. Instead of looking for little
green men, grid computing provides enterprises the means to utilize existing
unused processing power, memory, and bandwidth to perform large-scale
computational tasks that would otherwise require dedicated resources which
themselves might be largely underutilized a good part of the time.
While commercial grid computing solutions have to date been largely targeted at large enterprises, we think the general concept of allowing a business to effectively cluster its entire IT deployment for specific tasks is one that will not stay long in the rarified air of corporate datacenters. We see a host of technologies — including blade servers, high-performance computing, and clustering technologies — that are moving downstream into middle tier environments. We see no reason to think that Grid computing will not follow a similar path, as the medium-sized enterprise has as much to gain from leveraging and extending its existing IT deployment as the larger enterprise. IBM’s decision to include Grid capabilities in its new WebSphere Application Server will only help expand grid computing’s exposure and reach into the market, which means it will have more opportunity to move up and down the food chain as market conditions demand. For enterprises of all sizes, the ability to tap into a large pool of unused computing power may not result in the discovery of little green men, but it will allow those enterprises to capture some of the type of green that drops straight to the bottom line, a close encounter that most would welcome with open arms.
HP announced this week that it intends to submit its Web
Services Management Framework to the Organization for the Advancement of
Structured Information Standards (OASIS). Supporting the submission were a
number of middleware vendors including BEA Systems, Informatica, IONA, Sun
Microsystems, TIBCO Software, and webMethods. The specification outlines how
various network events can be monitored and communicated to management consoles
and describes how IT resources can expose management information about
themselves and how they can be managed. The specification largely focuses on
various elements of XML, which is used for information sharing across a range
of different system environments. HP said it will make the framework
specification available to the public for download. The company said the spec
was model and platform neutral, and supported J2EE and .NET, among others.
We’ve scratched our heads a bit over this announcement,
not because we think HP is making a mistake here, but in searching for the
downside to the company submitting this specification for standardization. In
short we don’t see any. For HP, an open standard framework for its management
tools — namely OpenView — gives the company back a
lot more than it might be giving up. Standards-based offerings expand
opportunities for ISVs and partners, thereby expanding the available number of
vendors providing solutions to IT customers. If those ISVs and supporters
decide to offer their wares on HP products, well, all the better.
We also see this move as one way that HP can maneuver around its lack of home-grown middleware solutions. The standardization process ensures — at least to a much greater degree — that various middleware products from vendors like BEA, Sun, TIBCO, webMethods and others will have some level of interoperability with HP products. In this way HP gets to have a bit of cake and eat it too, as the company does not have to go into the middleware market yet can tell its customers and channel that by deploying HP products, they will not be subjected to the modern horror of vendor lock-in. This leaves it up to middleware vendors to keep their products competitive, hopefully driving greater value to the customer base through active and engaged competition. Smart move, and one that should reap HP real benefits.
Pressure Drop: Sun Axes LX50 Entry
Level Server
Sun Microsystems has announced the end of life for the
company’s LX50 server. Originally introduced in August 2002, the Sun LX50 was
Sun’s first foray into the entry level x86-based server market. Priced starting
at $2,795, the LX50 could support either Solaris x86 or a Sun-developed Linux
OS (which the company recently discontinued). The LX50 also included the
company’s SunONE software stack, as well as a host of
other Sun applications.
A tendency to indulge in supercilious tea-leaf reading and
other related mumbo jumbo is part and parcel of an IT analyst’s life. Common
sense, experience, and imagination are critical to the job, but these are also
valuable qualities for the writing of popular fiction. That said, what are we to make of Sun’s decision to drop a product on
which the company spent considerable time and money? An obvious inference is
that the failure of the LX50 may be emblematic of the company’s overall
difficulties with the entry level IT market, most of whose users are devoted to
the x86-based platform, an architecture Sun has traditionally regarded as
little more than a plaything. Since the LX50 was also Sun’s initial commercial
Linux play, its cancellation might also denote continuing problems in Sun’s
stumbling participation in the Linux market. Such analyses may be fundamentally
accurate, but what the company should do to correct these perceived
shortcomings opens new areas of speculation. The fact is that Sun rose to prominence
and success as a take-no-prisoners devotee of 64-bit UNIX computing, a sector
the company continues to dominate. But the greater market appears to have
shifted since the dotcom salad days, and now seems to favor systems vendors who
offer a bit of something for every business IT taste. It is true that a
continuing recession and the remarkable evolution of x86 capabilities have
driven 32-bit solutions further into high-end computing than many ever
expected. It is also true that industry standard components are helping to
drive down server costs and profit margins in much the same way they have PCs.
Does Sun need to play in each and every IT field to remain successful? Many,
and the conventional wisdom, would insist that they must.
At times like this, it is wise to remember the ongoing trials of SGI (aka Silicon Graphics). Though it was the preeminent player among high-end graphics vendors, conventional wisdom suggested that SGI would be overtaken and left in the dust by upstart Windows-based graphics technologies. SGI’s decision to unload many of its visualization crown jewels and reposition itself as an NT server vendor might have seemed smart at the time, but excessive development costs and myriad lost opportunities spelled the end of the company’s role as a major IT player. Half a decade down the road, SGI remains a premiere, if much smaller, high-end graphics technology provider. At the same time those scary upstart graphics technologies were evolving, so it seems was SGI. Sun may appear to be confronted by insurmountable obstacles, but things change daily. Sometimes the wisest course of action is to lower expectations, hunker down, minimize your risks, ignore conventional wisdom, and continue doing what you know and do best.