Juan Carlos Perez

Subscribe to Juan Carlos Perez: eMailAlertsEmail Alerts
Get Juan Carlos Perez: homepageHomepage mobileMobile rssRSS facebookFacebook twitterTwitter linkedinLinkedIn

Related Topics: Virtualization Magazine

Virtualization: Article

Why an SCO win is a slam dunk and why you need not care

Just so you don't think LinuxWorld.com is of one mind about the SCO v. IBM case

(LinuxWorld) — When SCO filed its lawsuit against IBM, Linuxworld.com editor Mark Cappel summarized the complaint as:

SCO claims IBM is destroying the Unix market by taking knowledge the company gained via its source-code license to Unix and sharing this knowledge with the Linux community. This alleged action strengthens Linux, and, because Linux is no-cost or almost-no-cost, it cuts the legs out from under SCO's market. SCO claims IBM's duplicity has damaged SCO to the tune of $1 billion.

SCO's own summary, presumably drafted by someone at the law firm of Boies, Schiller & Flexner representing SCO, said much the same thing, but differently:

As set forth in more detail below, IBM has breached its own obligations to SCO, induced and encouraged others to breach their obligations to SCO, interfered with SCO's business, and engaged in unfair competition with SCO, including by
  1. misusing and misappropriating SCO's proprietary software;
  2. inducing, encouraging, and enabling others to misuse and misappropriate SCO's proprietary software; and
  3. incorporating (and inducing, encouraging, and enabling others to incorporate) SCO's proprietary software into open source software offerings.

I'm not a lawyer, but my understanding is that contracts lawsuits focus on promises and agreements. As Jay Feinman says in the book Law 101, "When someone make a promise and fails to keep it, contract law makes her pay because she has failed to bring about a future state of affairs to which her promise committed her."

In this particular case, the basic wrong alleged by SCO is that IBM licensed proprietary Unix information for use in AIX, promised to keep it secret, and then released that information improperly to the Linux community.

Being right isn't the point
As various people, most recently including Eric Raymond and Rob Landley at the Free Software Foundation, have shown, much of what SCO is made to say in the initial filing is either wrong or highly questionable.

My vision of this is of a clueless junior at the law firm being told to draft something and no one knowledgeable reviewing the result.

The first question is whether the errors matter because all of this stuff, along with most of the PR barrage, is probably just pre-negotiation posturing.

The second question is more interesting: if indeed this stuff was written by someone at the law firm and not formally approved by SCOsource, can SCO sue its own law firm over the misrepresentations in this document? (Editors's note: As with many legal questions, the answer is, "It depends.")

Most of the work on IBM's official project to port Linux to the System 390 took place at the Bvblingen skunk works (IBM Deutschland Entwicklung GmbH, Schvnaicherstr. 220. Bvblingen) in Germany where people also worked on AIX and other IBM Unix related projects.

Do a Google search for "IBM AIX Linux Boeblingen" and you get ten pages of results, including quite a few that seem to include resume information for people who worked on, or were influential with respect to, both Linux and AIX at IBM's Boeblingen center.

Given this apparent co-mingling of AIX and System/390 staff and expertise, I find it virtually impossible to believe that local expertise on Unix for the PowerPC architectures didn't influence the Linux port in areas where the problems they encountered had previously been addressed in work on AIX. Those areas would naturally have included nearly all the hardware/software interface issues characterizing significant IBM contributions to Linux for non-Intel environments like the System 390 and the AS/400.

For example, work done on:

  • Device drivers
  • Interrupt handlers
  • Floating point management (both software emulated and IEEE conformant)
  • SMP support and CPU virtualization
  • Memory management and system call handling
  • EBCDIC/ASCII tables and conversion routines
could all have benefited from such knowledge or code exchanges.

It is my expectation, therefore, that SCO will have little difficulty proving that IBM's Linux ports for its Z, I, and P Series machines were heavily influenced by its staff's knowledge of, and access to, the AT&T Unix source now owned by SCO.

If so, the route to general release of this code to the Linux community would most likely lie through SuSE first and Red Hat second, as these two firms were the first major players to integrate support for IBM's non-x86 Linux into their source code libraries.

However, proof that IBM exceeded the scope of its Unix licenses will, along with 50¢, get Chris Sontag (President of SCOsource) little more than guilt over harming IBM and a fine cup of coffee. To succeed financially with this lawsuit his lawyer, apparently the same David Boies who represented the Justice Department in the case against Microsoft, will have to do here what he failed to do there: prove the damages phase.

The Microsoft Unix License
Microsoft's first operating system offering (in 1980) was called Xenix — and was developed in partnership with SCO under an AT&T source license.

When SCO got the right to sell Xenix separately (as Unix) they agreed to license Microsoft's contributions to its development and so paid royalties to Microsoft until at least 1992 when they finally got the last Xenix code out of OpenServer. Three years later, SCO bought the Unix source business from Novell — and found itself issuing Microsoft's AT&T source license.

The creation of SCOsource with the mandate to collect on royalties due from about 30,000 Unix source licensees represents nothing more than SCO's somewhat belated recognition of the residual value in those licenses.

Thus Microsoft's public compliance with SCO's request that they renew their license is opportunistic, but not sinister. Sure they're seizing an opportunity to throw a little FUD into the Linux community, but really it's just business as usual.

To do that Boies has to prove first the existance of a valid, legally enforceable, binding agreement. That's easy. Second, Boies needs to prove that IBM breached the agreement in a material manner, and thirdly that this resulted in harm to SCO. I expect proving that the breach occured will be easy, but proving that harm resulted from it will be challenging even if SCO uncovers an internal IBM e-mail or two warning of concerns about license infringement. It's my belief, therefore, that since both sides understand this it is more likely this lawsuit will end through negotiation than in a judgment for or against SCO.

If so, this would explain why neither side wants to give anything away in public filings and therefore why, as Mark Cappel showed in his review of the case, SCO's individual supporting averments don't stand up to scrutiny — and neither does what SCO's Chris Sontag called IBM's "vanilla lawschool 101-type response."

What this means for Linux enterprises

If we assume that SCO will have little trouble proving the wrong but a great deal of difficulty proving consequences, what then is the bottom line for the general Linux community?

It is true SuSE and Red Hat probably have decisions to make in terms of trying to work with SCO, but there don't seem to be any significant legal issues for the Linux community in general.

Columbia Law School Professor Eben Moglen is a lawyer with special expertise in this area who acts as the pro bono publico general counsel for the Free Software Foundation. Here's what he told Thor Olavsrud for an internetnews.com story:

"There is absolute difficulty with this line of argument which ought to make everybody in the world aware that the letters that SCO has put out can be safely put in the wastebasket," Moglen told internetnews.com, noting that SCO distributed its own version of Linux with a kernel that allegedly contains Unix-derived code.

"From the moment that SCO distributed that code under the GNU General Public License, they would have given everybody in the world the right to copy, modify and distribute that code freely," he said. "From the moment SCO distributed the Linux kernel under GPL, they licensed the use. Always. That's what our license says."

In contrast, there are both legal and practical issues for IBM that could have far-reaching consequences.

First among those is SCO's willingness to enforce its contract with respect to AIX. The 100-days notice to IBM required under that contract expires on Friday, June 13th. If there's no settlement or continuance before that, then anyone directly or indirectly paying IBM for AIX should probably seek legal advice before making that check payable to IBM.

Perhaps killing mainframe Linux would be the kindest thing
There's a new IBM Redbook on Performance Tuning Linux for the zSeries that includes a lot of IBM certified performance information.

This isn't a benchmark book but in the course of tuning work on a dual-IFL z900 they reveal lots of performance information — much of which you won't believe on first reading.

For example:

  1. Turning off daemons like cron is an important tuning step because these waste valuable system resources.
  2. Using a RAM disk with an experimental driver and a bunch of kernel hacks pays off on speeding up paging — all the way to about 40-megabytes per second;
  3. On real drives using striping and multiple channels pays off big time. With only eight disks and four fiber channels, the machine delivers a blistering 120-megabytes per second.

On the positive side, rumor has it that IBM is in an overstock position on z800s so all those Intel 80386 users looking to upgrade to the mainframe can probably get a pretty good deal — say half a million all in for a four-CPU box?

More interestingly, IBM may find it easier to use the courts to continue its contracts with respect to AIX past June 13th than to continue supporting Linux for its non-Intel-based products. If so they they could either negotiate a quick settlement with SCO or defy SCO to continue selling the Linux products while dragging out the matter in court.

In the latter case, IBM would have to assure its customers of its willingness to protect them from SCO legal action. However, SCO needs to win only one judgment against an IBM customer anywhere in the world to create enough FUD to bring that strategy to a crashing halt — a consideration that I think probably motivated the otherwise rather idiotic letter SCO recently sent to 1,500 or so presumtive Linux users.

If IBM decides to fight rather than settle, it will face enormous risks and need to put a Plan B in place just in case SCO manages to get a court ruling before it runs out of cash. Remember, to get a court order enforcing its cease-and-desist order on AIX, all SCO has to show is that a material breach of the contract occurred. Consequences matter in terms of compensation, not in matters of fact.

If SCO wins its order, IBM's ability to sell Linux on its Intel-based machines will probably be largely unaffected, but pSeries sales would stop and there would be serious questions surrounding the Unix System Services components within zOS and OS/400. That's an enormous risk, and IBM's choice of a Plan B to offset it correspondingly revealing about about IBM's own beliefs about its vulnerability.

None of the choices I've been able to think of seem very attractive. For example, IBM could:

  1. Buy a Unix-like operating system, perhaps by taking a run at Sun or by doing a deal with HP to take over the Alpha and Tru64;
  2. Buy or rent all or part of Novell's assets to gain ownership of the copyrights and patents underlying their contract with SCOsource and hope to sow enough confusion to stretch the legal issue beyond SCO's means;
  3. Restart its Linux port using the Debian code base and accelerate the move to phase out AIX in favor of Linux — while either re-inventing the S/390 port or restarting from the original (and probably uncontaminated) Linux for 370 project run by Linas Vepstas;
  4. Recognize Linux is just politically correct Unix, they could switch everything to BSD and rely on their size in the market to make it stick.

Since these are all bad choices, I'm sure IBM will either settle, enter into serious negotiations and thus get SCO to lift the deadline, ask a court for a temporary injunction, or come up with a better overall answer.

Meanwhile the general Linux community should probably treat this as a spectator sport: Sitting back to see what happens on June 13th.

More Stories By Paul Murphy

Paul Murphy wrote and published 'The Unix Guide to Defenestration'. Murphy is a 20-year veteran of the IT consulting industry.

Comments (1) View Comments

Share your thoughts on this story.

Add your comment
You must be signed in to add a comment. Sign-in | Register

In accordance with our Comment Policy, we encourage comments that are on topic, relevant and to-the-point. We will remove comments that include profanity, personal attacks, racial slurs, threats of violence, or other inappropriate material that violates our Terms and Conditions, and will block users who make repeated violations. We ask all readers to expect diversity of opinion and to treat one another with dignity and respect.

Most Recent Comments
Gray 09/19/07 07:19:00 AM EDT

Well, I wonder how the author is thinking about this article now, in retrospect. Yeah, life's a bitch. It always will turn out different than you thought...

No misunderstanding pls, this what a quite thoughtful column for it's time, but not a single prediction became true!