Big Guns Unite To Unify Unix

MikeDartt writes "Wired reports that Compaq has just joined IBM and SCO in Project Monterey, which is an attempt to get a single UNIX distro that will run on Merced. Perhaps I'm naive, but why get behind a new *NIX as well as Linux, esp. when the latter is both more open and more fashionable? "

Comments

What's up with Hurd?

[jerodd]

Hurd is alive and well. Hurd addresses the high-end issues; Hurd has been designed from the ground up to handle gargantuan (dozen terabyte) files and incredibly high workloads and scale to dozens of CPUs. You can read more about over at the the GNU project website. If you'd like to toy with Hurd, then go download Hurd kernel from the aforementioned website and then get Debian GNU/Hurd, which supplies the user-level operating system environment. It's quite nice, albeit unstable.

Cheers,
Joshua.

but it is more than just that

[jerodd]

By using Linux's RAID features you can achieve the same results as with LVM, so I don't see what LVM is offering. There is no slick GUI to administer the RAID, but a few hours spent alone with Tk could come up with something nice to add disks, rearrange partitions, etc.

What we need is the ext3 filesystem. This will support, among other things, integral distribution across volumes (which is something the LVM does quite nicely), resizing of partitions, possibly while the system is running (to add new fixed disks on the fly, as I am wont to do), and journalling. Journalling will banish the dreaded hour-long fsck from computing forever more. (The JFS is so good I use it exclusively when I work on XFree86 in OS/2, because when the Xserver kills my OS/2 box, I just poke the power and the UJFS.DLL CHKDSK takes about five seconds to run, as opposed to the five minutes HPFS CHKDSK took.)

I need to get involved with the ext3 project. I've got too many things to do right now, but I need to learn serious kernel hacking one day here.

Cheers,
Joshua.

Monterey is nothing to fear

[PedXing]

This project is NOT a plan to unify UNIX in general, but rather a plan to make a new UNIX for IA-64.

SCO has experience with x86 (and owns UNIX) and long ago announced that they will port it to IA-64. Now Compaq and IBM (and Sequent) announce that they will help with SCO's UNIX for the platform and ship it on their IA-64 servers. IBM also plans to port it to their RISC hardware. Compaq sees Monterey as a way to move Digital UNIX (or whatever they call it this month) forward.

In other words, Compaq and IBM don't want to spend a huge amount of money to port their UNIXes to IA-64 to compete with everyone else on that platform. If it works out, then they have a foot in the door. Otherwise, they can just dump it and move on. This is NOT a vote of confidence in IA-64's future!

Sun announced that they will port Solaris x86 to IA-64 and Fujitsu and NCR (if I remember correctly) jumped on their ship. HP will, of course, have HP-UX on IA-64. And Linux is also in development for the platform.

I'm sure Compaq, IBM, and everyone else will ship Linux pre-installed on their IA-64 servers when they come along. They're just covering the "proprietary UNIX" base at the same time in the cheapest possible way.

Note that I'm saying IA-64, not Merced. I've long believed that Merced would never ship in volume because of production delays and poor performance. Only the second-generation IA-64 part will have a chance of success. I'll stand by that prediction...

more open and more fashionable?

[kvajk]

Linux is great. But enough dogma; it's not a high-end OS, and it won't be for quite some time.

The fact that it's the best OS for your PC today does NOT mean it's the best OS for an enterprise server today. Open source will eventually move into that space, I'm sure, but it isn't in there yet. (And no, a quad xeon isn't even close to high-end.)

Saying "why not Linux?" when it's perfectly obvious why not only serves to make Linux, a great OS, look like little more than a bunch of hype.

There are some lawsuits against SW vendors

[alkali]

This means that even if the product doesn't do a single thing it claims to be able to do, THE USER HAS NO LEGAL GROUNDS FOR SUIT!

I don't think this is exactly right: If the software vendor had absolutely no obligations whatsoever, the contract between the vendor and customer might be considered "unsupported by consideration" (ah, legal jargon) and therefore unenforceable as a contract.

More common, I think, are limitation of liability clauses which state that the vendor's liability is limited to a refund of the purchase price of the software license. Such limitations are common in many other industries as well, and are regularly enforced by courts.

[You might feel that such limitations are unconscionable, and should therefore be ignored by the courts. The standard reply is while courts can readily tell whether a contract is supported by some valuable consideration on both sides, it's not a court's job to evaluate whether a contract is a good deal -- it's the market's. If you don't like the terms of the software license, you should find another software vendor. That's the argument, anyway.]

Incidentally, contrary to evilpenguin's comment, it is not at all uncommon for businesses to sue their software vendors; in particular, many vendors have been (or anticipate being) sued as a result of the Y2K problem. Read a recent 10-Q or 10-K from any major software vendor for a discussion of this issue. It's relatively rare, I think, for ordinary consumers to sue software vendors, but you can read about one such case at the web site of the Milberg Weiss law firm (search with the keyword Issokson -- it's a Y2K case).

why not just unify behind linux?

[FWMiller]

There are a number of reasons why UNIX vendors would not be unifying behind Linux as the "unified" UNIX.

1) These vendors have invested huge amounts of development effort in putting together their own implementations. They will not just walk away from their own design decisions because of Linux's rising market share.

2) Despite the increasing popularity of Linux, newer, high-performance commodity and custom hardware devices are driven by proprietary UNIX implementations first. This may be changing but it still has a ways to go before it changes (if it ever does IMHO).

3) Culturally, Linux is NIH (Not Invented Here). This may sound silly but its very pervasive and there are some convincing arguments why it will continue. The developers and customers for a given UNIX, even it they are closet Linux hackers, will take product to market with an in-house solution over Linux first because it has a critical mass within the organization that allows faster time to market.

4) Linux may incorporate many very modern OS features and implementation designs but other UNIX vendors will always believe, justified or not, they have a superior design. In such a case it may incumbent on Linux to adapt to their design rather than the other way around.

All of these factors may wane over time, but I doubt they will disappear completely. The short history of operating systems has never seen a single UNIX spec and I'm unconvinced it will happen now.


Later,
FM