Oracle Creates Linux Division
MikeDartt writes "According to C|Net news, Oracle is devoting an entire division to creating a line of products for Linux. As the article notes, this isn't surprising, given Ellison's antipathy towards MS. Yet another announcement for LWE, eh? " Will this lead to even more enterprise credibility?
I received an early adopters CD of Oracle 8i on linux, and I have to say WTF Oracle? The installation uses a Java based installer, hard coded to run from /usr/local/jre ... Hmmm nothin in the docs about chaning that. Ok, create a symbolic link... hmmm Can't run the install as root? Ok... then during the installation I have to drop to shell and run some root enabling script... cant two temp dirs because of privilages? Hmmmm...drop to shell, su to root, create the dirs by hand...back to the installer... then it tries to create a directory in /proc and fails... finally it gives me an error saying Java.Lang.Thread failed... Great testing guys, and Im runnin Redhat 6...The html docs are horrible, and mention none of this, plus the installer gives me no clues as to what and how to name my database and associated accounts. Then when all is installed and done, I'm left with a bunch of files in a OraHome1/bin dir that I have no idea what they do. How do I run it? How do I administer it? SQL 7 is feeling pretty comfy right now... -doog
I am just trying to point out that popularity some times flies in the face of logic. If we have to go with Linux, then, the LSB will be more important than ever... And, along those lines, it would mean that the LSB would be more important than standards getting in the way of development of Linux itself, because without standards, someone like Red Hat could end up being the only Linux that VAR/ISV's want anything to do with.
Since the synopsis mentions making Linux more credible in enterprise environments, there's one enterprise-level feature that wasn't in Linux the last time I looked (which, admittedly, was early 1998): >16bit UIDs. Does any distro currently support more than 65536 users?
[PLEASE don't flame me for not knowing this.]
I wouldn't be a bit surprised if Ellison decides to sell complete servers with an Oracle Linux distribution installed. They'd probably put an emphasis on speed and security and integrate an oracle database server.
Then, they'd start making all kinds of 'improvements' to non-kernel-related aspects of the Oracle Linux distribution.
Vidi, Vici, Veni
Truthfully, there are a few more areas where Linux isn't ready for the enterprise. At least for mission critical applications.
Primarily the missing elements relate to high availability applications, with failover capability.
Things like a journalling file system, and HA clustering (no, Beowulf doesn't do high-availability) are desperately needed. I know there are some early projects in the works to address these issues, but they don't seem to be ready for prime-time just yet.
Now, if I could run Oracle 8i Enterprise for Linux with the parallel server and hot standby database options enabled, I would be happy. But, truthfully, when you get a web license for Oracle (licensed by CPU power, not per concurrent user), the costs of the Oracle licenses are so much higher than hardware/os licensing costs, that buying a couple of Ultras and SunCluster isn't that big of a deal anymore. What's $100,000 worth of hardware/os when the Oracle licenses cost $500,000.
And, if you don't need the high availability features of 8i Enterprise, then you probably don't need the Enterprise version at all. Then, Oracle 8 on Linux is well suited for your needs. At least that's what I've found to be true.
Linux is evolving rapidly, and I do prefer it for ordinary work, but it just isn't in the game yet when it comes to enterprise level, high availability implementations.
Ellison may have antipathy towards MS, but I'll wager that has little to do with the decision. Ellison is a businessman and antipathy earns little money by itself. Ellison won't start a linux division unless he knows it will make money. (He might enjoy a linux division more if it succeeds, contributing to the success of linux and hurting Windoze, but that's another matter.)
--JT
Don't think so. Larry left his company to dwindle and disintegrate amid rabid infighting while he sailed around the world on his 80 foot yacht. In recent years he's been acting like managing oracle is a hobby he indulges in from time to time. Oracle has suffered from this. I doubt Larry, or anyone else at Oracle believes that they selling a ridiculously overpriced piece of software that runs on a free OS is sound buisness. They just don't want to get left behind in the great linux stampede.
--Shoeboy
IIRC, the Raw Iron initiative involves a minimal Solaris kernel without the usual trappings of the OS. Administrators would scarcely know Solaris was there providing low level services to the Oracle software. Presumably, the installation disks for Raw Iron would include the minimal OS so you could literaly start with a "raw iron" computer and a CD and do a single install process.
After installation, the advantage would be to allow Oracle database analysts to do the complete administration of a dedicated database box without support from a sysadmin. Of course, it could be tuned specifically for good database performance without considering other services -- other than SQL*Net type things that support database connectivity.
Geeky modern art T-shirts
I understand Oracle has been ported only to x86 Linux. But 32 bit architecture is quickly running out of steam. In fact, for large databases 4 gig memory space is nowhere near enough already. I wonder if they plan to port it to other platforms, such as Alpha or PPC. Insidentally, how well does Linux support 64-bit memory on these platforms? What about SMP on them?
___
If you think big enough, you'll never have to do it.
Maybe I shouldn't have put 'improvements' in quotes like that. I honestly don't care how many/few distros there are, or who makes them.
I was just putting a prediction out there in case, in the unlikely case that I am right, I can say "I knew it" to nobody in particular.
It just struck me as the type of thing Oracle would do. I am actually VERY happy every time some company moves away from windows toward Linux, not out of any particular wish to see Bill go down in flames, but because then the chances are just that much better that I will use Linux at work.
I hate using windows at work. grrrrrrrrrrr....
Vidi, Vici, Veni
Now, on workstations, I am still using Linux, because of the better avaliability of software. But, for something like Oracle, I would prefer a direct port to FreeBSD if I were to ever use it. Because of the fact that Oracle has the potential to use a lot of memory, FreeBSD may have an edge in better memory management (always lower memory usage, IMHO it does a more efficent job freeing up used memory while caching the important stuff).
Heh.. Port and ports. I mean a commercial binary of Oracle avaliable for FreeBSD when I mention a "direct port to FreeBSD," not to be confuzed with the FreeBSD ported software that can be compiled from "/usr/ports" (and also not to be confuzed with the type of port, like 80, that everyone is trying to telnet to for the crack.linuxppc.org thing).