Darl McBride, head of Caldera nee SCO, has an open letter proudly begging the world to keep taking his company seriously. As a member of the Open Source community, and also of the larger community of all Unices (having used a number of Unices before Linux, and still liking a lot of them), I'll make what is one of many responses, in typical usenet fashion. Note that paragraphs are omitted when I don't have anything to say about them, also in Usenet tradition.
Long Live Unix
Three years ago, when I first joined The SCO Group, we focused the company on the area that was most profitable and provided the most benefit to customers, investors, resellers, developers and employees: UNIX. People thought we were crazy. But since SCO owns the UNIX operating system and it made up 95 percent of our company's revenue, and we were getting strong demand from customers for a next generation version of UNIX, that's where we concentrated our efforts.
Except it wasn't pretending to be the original SCO at the time -- it was still going by the more honest name of Caldera. Before Darl joined, Caldera was a software company that primarily made a Linux distribution, and maintained the ancient, decrepit Unices from the original SCO, SCO Openserver 5.x and Novell Unixware 7. These were correctly understood as legacy platforms, with no large-scale development planned and maintenance continuing at a low key to keep alive people who had not yet migrated to a modern platform for one reason or another. One company I was a sysadmin for several years ago was a good example of a company that was only very slowly migrating to a modern OS, because a lot of development was tied to both of SCO's platforms. If our codebase hadn't been such a nonportable mess (I was working on refactoring it too), we probably would've stopped a slow migration to Unixware 7 and migrated to Linux much more quickly. As is, because we were delivering closed-box solutions to customers, we were stuck with SCO in the short to medium term. Anything new we did that wasn't tied to that architecture was tied either to Solaris/SPARC or Linux/x86. Anyhow, also at that time, Caldera was talking about an upgrade to Unixware called OpenUnix, which would be a more modern Unixware 7 with the ability to run either a Linux kernel or its own (Unixware family) kernel, modern Java, and a bunch of other things that Linux and modern Unices have that neither Unixware nor Openserver have. Post-Darl, Caldera renamed itself to "The SCO Group" (wannabe-ism of the Santa Cruz Organization), focused on the old, crufty stuff, and began suing everyone. This is a case of them setting fire to their neighbour's barn -- they products were inferior to Linux, so they decided to attempt to kill Linux in business and focus on restoring the pre-Linux OS landscape. Unfortunately for them, even if they had managed to do so, neither the Unixware nor the Openserver lines that they had bought were even close to best-of-breed. With Linux out, Solaris/x86 would be the most obvious replacement in businesses for Linux (FreeBSD coming a close second). A world without Linux would not help SCOGroup very much. The strong demand from his customers for a next generation Unix is probably charitable at best -- his customers don't particularly want to be his customers, they're just not quite ready to leave yet. Even if he were interested in making an updated bloodline Unix, he would've continued the Unixware/OpenUNIX line -- Openserver is more ancient and more of a dead end than Unixware. It does say something that shortly after Darl came on board, the founder of Caldera, Ransom Love, resigned in disgust.
One thing is certain: while our lawyers are protecting UNIX in the courtroom, SCO is clearly focused on winning in the marketplace with superior technology and better value for our customers.
In June, we released SCO OpenServer 6, which was a multi-year, multi-million dollar development effort that resulted in a product that goes beyond simply leveling the playing field with Linux. Based on the feedback from our strategic partners, customers, resellers, engineers, and many others, I believe SCO OpenServer 6 outshines Linux on a number of fronts
Fortunately, his lawyers are failing in the courtroom. It is telling that he has essentially bet the company on the lawsuits that he's been using to attempt to kill Linux, and it is suspected that if they fail, there will be nothing left of SCOGroup. When he started the legal action, he let go of much of the product development staff of SCOGroup, so I doubt it's very different than a hybrid of OpenServer 5.0.6 and OpenUNIX 8.x. In the last releases of OpenServer 5, they made a big deal about being better than their competition when all they did was take some popular Linux packages and compile them for OpenServer. This is no major achievement -- when I was maintaining SCO7 and SCO5 years ago, I got the same packages, just downloading them from a third party repackaging site. Let's take a look though. Here is a link to a PDF with up-to-date info on SCO6 versus SCO5. It looks like they actually used the OpenUnix kernel as a base, and added a few new things. They moved from X11R5 (ancient) to X.org's X server, catching up with Linux on that front (using open source code to do so). Their users are still stuck on Apache 1.x and classic Mozilla (lucky them). They compiled CUPS, also adding some Linux stuff to catch up.. and an old version of Java. They also compiled some opensource database software, and .. hmm. That's about it over OpenUnix. Yup, SCO6 adds very little new to OpenUnix 8, and where it does, it just compiled some open source stuff to play catch-up. No surprise.
OpenServer 6 Costs Less - OpenServer 6 offers very aggressive pricing. The purchase price for SCO OpenServer 6 is priced from $599 to $1399 which includes the license to the product, software fixes, and access to SCO's online knowledge base. Customers pay once for the product and run it for as long as they like.
Linux vendors, on the other hand, seem to have a "bait and switch" pricing model. The initial attraction to Linux was a price tag of zero cost. Yet, they typically charge customers from $349 to $2,499 every single year. Calculating the cost of running Linux over a five year period of time, that same customer pays from $1,745 to $12,495. Since the Linux license itself is "free", are you really happy to be paying annual subscription fees that are, in effect, higher than SCO's price for both licenses and software fixes?
It looks like Darl is looking specifically at Redhat's price list for Redhat Enterprise Linux. Let's also see SuSE. He's (sort of) telling the truth here, but not the whole truth. With Redhat or SuSE, you do subscribe to it rather than buy it, but that is for support. Users of either system can buy it for a year and then cancel the subscription, and continue to run it while only getting security patches. With SCOGroup's Unix, if you buy it, you get the product and a certain level of support for awhile (access to most patches) until they eventually stop making many patches for your version, at which point you pay again to get a new version. You might additionally pay to get access to certain patches -- a number of them are unavailable without more money. In this way, it's hard to compare the prices because with SCOGroup, the product and support slowly peter out, and if you don't want that to happen, you'll do things in the continuing cost model (like Redhat is by default) of getting each release as it becomes available. This is also discounting the alternative pricing model that a lot of groups go with -- when an organization gets large enough, they often instead get some really skilled sysadmins and go with a free Linux which they then have said sysadmins support entirely, without outside commercial help. As usual, as is the case with Microsoft, SCOGroup's knowledge base isn't particularly useful.
SCO Has a Superior Kernel - SCO OpenServer 6 includes the UNIX System V Release 5 (SVR5) kernel, the result of more than 25 years of high-end development work that has created a proven track record of stability and reliability. With our latest release, OpenServer provides support for up to 32 processors, 64 GB of memory, terabyte file sizes, and full support for multi-threaded applications. Linux is still young from an operating system perspective. I would challenge any kernel out there to match us head-to-head. While Linux may appeal to some as the sleek, new "racer" on the track, the experienced IT professional will truly see the real power under the hood when they test the UNIX kernel and the tried and true power of UNIX combined with the new capabilities of SCO OpenServer 6.
This is hilarious. SVR5 had nothing to do with old SCO, which dealt only with ancient versions of Unix, not the SVR4 stuff whichwas largely done by Sun. The only thing that let SCOSource create the misleading name of SVR5 is their purchase of some trademarks from a few other companies. None of the stuff they brag about above is new, nor is it something they have much experience with, SCO6/OpenUnix being very young and the only products they have made that offer this stuff. As for the experienced IT professional, practically none of them are or have been looking at SCO, old or new, for several years. Further, Linux is a Unix. Unix may be a trademark, but common usage of the term includes BSD (which is bloodline Unix, kind of) and Linux (which is not bloodline, but nontheless fits in the category). The Linux and Solaris kernels (and OS's in general) are both more capable and proven than anything SCOGroup (or old SCO) has ever produced. Old SCO was always near the back of the pack of Unices, and SCOGroup is likewise uninteresting and unlikable.
OpenServer Has Better Security - IT managers rank security today as one of the most important decision factors in selecting an operating system. According to technology risk management firm mi2g, SCO OpenServer is one of the most secure operating systems in the world. A study confirmed that SCO UNIX platforms had the lowest number of vulnerabilities of any operating system they had studied. SCO OpenServer 6 has all the latest security protocols and encryption systems.
mi2g? Who are they again? Let's ask google.. Hmm... Doesn't look like mi2g is particularly well-known or trusted. Ask google yourself.
Unfortunately for Linux, mi2g also confirmed that the Linux operating system has become somewhat of a hacker's paradise. In a study conducted only seven months ago they found that overall, the most vulnerable operating system for manual hacker attacks was Linux, accounting for 65.64% of all hacker breaches reported.
That's an interesting way to calculate vulnerability. After all, we wouldn't expect OS's that are installed more to be hacked more, would we? We would also not expect systems that are more useful when hacked to be hacked more, or machines more likely to be on the net to be hacked more... After all, equal interest in getting into the few SCOGroup systems running out there is going to cause all of those systems to get hacked, whether they're on the internet or not, as often as the many more systems running Linux or windows.
SCO Has a Customer-Driven Roadmap - Customers expect to see a published roadmap of product development. Linux development plans and schedules are generally as unknown as they are unpredictable. Contrary to that approach, SCO believes in a solid, public, and planned roadmap based on the tried and true methodology of listening to customers, evaluating technology and bringing it to market in a timely manner. SCO is committed to deliver on its roadmap promises--on time and on target.
Linux will likely continue to face challenges about its development methodologies and roadmaps as long as it continues to be a loosely organized set of volunteers who develop what they want, when they want.
Does SCOGroup have a roadmap? After poking around on their site, and eventually using google to search their site, I wasn't able to find anything beyond mentioning that their traveling roadshows might mention one. By contrast, searching redhat.com for roadmap came up with roadmaps for a number of their projects, as well as a lot of communication between engineers and end users as to what's coming next. It actually looks like neither of them have much of a roadmap proper though.
OpenServer 6 is Backward Compatible - In listening to our customers, we've received the strong message that backward compatibility is essential. Backward compatibility is almost non-existent for Linux customers. Linux has a "community" of contributing volunteers, and while some would say this is a boon for Linux, I would characterize it as a bane because channeling all of these contributions into another point release for Linux inevitably causes problems. Who is checking for compatibility across thousands of applications, drivers, hardware and peripherals? Who is verifying backward compatibility? When a new upgrade of Linux is required, software vendors and end users most likely have to upgrade their application as well.
SCO OpenServer 6 customers get a stable operating system with full compatibility for applications back to the earliest versions of SCO OpenServer and Xenix. SCO customers don't worry that their application won't run with the new version of their SCO operating system because backward compatibility is built into each new release. It's part of the product release criteria, and SCO's focused engineering team makes it happen every time. As is the case with OpenServer 6, older applications written on this operating system work seamlessly with the new features and capabilities built into the product.
That's odd. I remember very occasional incompatibilities between Unixware and OpenServer. In theory, the binary compatibility stuff on Linux should be able to run older SCO binaries, although I'm not sure if anyone has cared for a long time. SCO binaries are pretty rare, and most stuff Unix systems run nowadays is source-available, with the rest commercially maintained and current. Have I ever been bitten by backwards compatibility on Linux? There were a few closed-source games I used to run that don't run anymore because they relied on some unpublished and unsupported bits of kernel threading code. That kinda sucks, but it's a very rare occurence (on SCOGroup or Linux). It's probably easier for SCOGroup than for the Linux folk though, because their OS's have almost never undergone serious changes. Openserver6 is essentially OpenUNIX 8, which is essentially a somewhat freshened Unixware 7. Openserver 5 was old and stagnant for a very very long time. Old SCO and SCOGroup never had to worry about backwards compatibility because they never really kept their systems from becoming up to date. It's like not worrying about the cost of getting fresh food by letting food live in the fridge for years.
SCO Allows You to Focus on Your Core Competency - A popular animation on the Internet features a guy named Steve, the Linux Super Villain. During the course of the 60 second animation, he describes his work with Linux stating, "First you have to config it, then write some shell scripts, update your RPMs, partition your drives, patch your kernel, compile your binaries and check your version dependencies." While the animation is designed to be humorous, it's not far from the truth. If you're adopting Linux, get prepared to go into the operating system business because that's exactly the path you will be taking.
One of the primary reasons customers choose SCO is that they don't want to be an operating system vendor. They want to be free to manage their businesses, and leave the operating system details to SCO and our army of resellers, support engineers, and product development personnel
You want that, it's an option. You don't want that, then don't go with one of the do-it-yourself distributions. People using RHEL, SuSE, or even Fedora don't do anything more than SCOGroup folk do. Of course, SCOGroup people also need to config it, possibly write shell scripts for the same reason Linux folk do, update their packages (they're not in RPM format in Unixware or Openserver), and partition drives (SCOAdmin and the distro-appropriate setup tools all make this easy though). With either, you might be compiling binaries, depending on what you're doing. SCOGroup isn't magic -- some things are naturally part of running a computer, and a lot of this stuff happens in a slightly different form even on Windows and MacOSX. The point isn't avoiding these tasks (although patching kernels doesn't happen so often anymore anywhere) -- it's making them intuitive and easy. Fedora and RHEL make this quite simple. SCOAdmin actually wasn't a terrible tool either for some of these things -- while the rest of Unixware and Openserver was all-around awful, SCOAdmin was actually ok.
SCO Owns and Warrantees its Products - SCO owns the OpenServer 6 operating system that it licenses to its customers. SCO also owns the UNIX operating system technology that has been licensed to thousands of firms over the years. Alternatively, Linux distributors ship an operating system for which they have little control and no ownership. In fact, the General Public License, which governs the use and distribution of the Linux operating system, makes it clear that Linux conveys no warranty to end users. From the standpoint of intellectual property rights, SCO OpenServer 6 is backed by a company that warrantees its products.
Why would Linux distributors' exclusive ownership over Linux be a good thing? That would make it hard for them to compete as freely. It might be good for SCOGroup, because then Linux development would stagnate and SCOGroup could point at particular flaws in a single vendor, but that isn't the case. Again, if you don't want warranties, you can go with a do it yourself distro. If you want warranties and support, you can get them.
SCO is Unifying its Code Base - Yogi Berra once said, "If you come to a fork in the road, take it." Forking is exactly what is happening to Linux. The grand promise of Linux was that it wouldn't fork or fragment into multiple Linux operating systems. A noble sentiment, to be sure; but Linux distributors have ensured exactly the opposite. They are attempting to get ISVs locked into a specific flavor of Linux, thereby forking Linux with every new version of the product
Grand promise? I don't remember anyone making that promise ever. In fact, it's almost always been seen as a good thing that there are lots of different distributions out there. There are groups out there like the LSB that set standards that Linuces generally stick to so application vendors can make software that runs on all compliant OS's, and that's been a pretty successful and good thing.
SCO UNIX: Legendary Reliability - Customers value and trust a vendor whose products provide reliability and stability year after year. A good operating system is like a strong building foundation, you may not think about it everyday, but you're glad it's there.
The problem is, I remember running Openserver and Unixware. Neither were particularly reliable, and we often had to get specially dated hardware in order to get them to run. Our Solaris and Linux boxen were much more reliable, and much better documented. Fairly often, the package system became corrupt on the SCO boxen, and when that happens, you might as well reinstall the operating system from scratch because there's no way to repair it (that I've ever seen on SCO's website or elsewhere). Maybe it was my fault for wanting to use the package system in the first place to download third-party packagings of useful things that should've been in the OS, like ssh.
SCO Has an Award-Winning Support Team - Customers of OpenServer 6 have access to a support team that knows the product inside and out. They have decades of experience with our product line and are available around the world and around the clock. In addition, the SCO support team has access to the very development engineers who created the product. This cannot be said of Linux distributions. For most customers who have an immediate need, SCO can respond much faster than Linux because our support staff is in-house and has direct access to the developers to answer all customer questions.
Conversely, when Linux customers run into problems and need professional technical support they really have only two choices. First, they can turn to the Linux distributor who played a big role in packaging the product but had nothing to do with its core development. Or second, they can turn to the Linux volunteer community. These volunteers were not paid to develop the product; and they received nothing from the Linux distributor, there's no obligation for that volunteer to support the product. Would you really want to trust the backbone of your business to the likely unpredictable response times of this Linux "volunteer fire department" support model?
Most large vendors are going to go with an OS vendor, and the OS vendors support all the software they ship. They generally have the source, after all, and are fully competent in finding problems and fixing issues. Guess what? SCOGroup DOES THE SAME THING. Note that they advertise including Apache, ssh, and other tools as part of their OS. These are open source products that they don't make themselves. How do they support them for their customers? They have the source, and if they need to, they can talk to the development teams for these products. The developers for big open software are often working for someone who finds their opensource product useful anyhow, like IBM or some Linux distribution vendor, so they're not that hard to contact.
One of the ways Darl gets tricky is that he bounces back and forth between the two primary ways of using Linux without mentioning either or their differences much, using whichever of them for comparison is most useful to bash. In the first scenario, a company goes through the OS vendor for most things, buying support, doing licenses, getting warranties and service level agreements, and things of that sort. Life in that sphere isn't very different between SCOGroup's products and using a Linux vendor like Redhat. In the second scenario, a company takes a DIY approach, going with a free Linux and getting their IT group to manage all the support needs. These represent two very different ways of approaching the problem, and provide very different things. It is nice to have the choice, and the fact that the second is possible means that there are a lot of techies outside of distro vendors who know Linux inside and out (like me), eventually meaning that we contribute to the community and make things better both for DIY folk and vendor folk. The traditional Unix sysadmin attitude was somewhere between vendor-dependence and sysadmin-dependence-- we have the source now, and we're better off because of it.
One additional area of relevancy is that SCOGroup's OS is an inferior product. Its generic tools are still badly outdated, the filesystem is still a mess, the desktop is dated, and the company has made enough of a nuisance of itself that nobody wants to deal with it as a software supplier or a customer. SCOGroup is basically puttering out, and Darl is trying to put an optimistic face on his folly.