Re: Oracle on Linux enhancement request, need your support!

["Pei L. Ku" <pku_at_gte.net>]
|
| * Assuming Linux commnuity is advocating for a cost-free Oracle on
| Linux, how are you going to convince Oracle to spend development
| money on a port that will essentially be a zero-revenue (or worse,
| negative earning) port? After all, Oracle is a proft-seeking
| business.

whatever gave you the impression that people are looking for a cost-free port of Oracle on Linux? the operating system may be free, but that doesn't mean everything that runs under it has to be free as well.

I too do a lot of development work with Oracle and other databases and if a native Linux port of Oracle was available I would order a license so fast the sales people wouldn't even know what hit them.

the reason I mainly use Linux for development (even when I'm at customer sites that run some other flavor of UNIX) is because Linux is the perfect environment to develop portable UNIX applications. you have a very complete set of excellent development tools available right out of the box and the OS helps you along nicely if you need to track down a bug or see what a process is doing.

an example of a database company that ported their database to Linux is Solid. because of their excellent portability, outstanding customer support and the fact that they take Linux very seriously as an OS I now use Solid for most applications that need a database. I even have a server installed on my portable so that I can develop software (under Linux) at home.

besides, Windows 95 is in practice free, since it is usually pre-installed when you buy a PC these days at no apparent (to the customer) extra cost. you don't see anyone thinking "oh, the OS is virtually free, so we'll better give away the software that is to run on it...".

| * Assuming the Linux community is willing to pay licensing fee for
| an Oracle port, issues surrounding tech support can still be sticky:
| If you see an ORA-600 error on Linux, who do you call? How do you
| know if it's an Oracle problem, a Linux problem (which
| flaver/version/patch of Linux), or a hardware issue?

the Linux community has a history of taking care of itself. having worked in a few large companies that have UNIX servers running Oracle it is *very* apparent to me that UNIX gurus are few and far between. these people call support and have someone fix their problems.

the Linux community is different. first of all, the community is more densely populated with people who have intimate knowledge of their systems. something breaks or stops working, chances are that the problem will be solved within hours or days -- if the person who detected the problem isn't able to resolve the problem herself.

Linux too is different from OSes like HP-UX or NT in the sense that it was made by programmers for programmers. isolating and explaining errors on a Linux system is a lot easier than under most other OSes because the OS itself and the tools that usually come with it can be used to probe almost every aspect of the system.

I've been working on commercial UNIXen running Oracle and the one common denominator is that I always keep wishing I could develop the code under Linux so I wouldn't have to spend time trying to iron out wrinkles with primitive programming tools.

| On the other hand, this could be a bonanza for Oracle if their Tech
| Support charges Linux users for tech support by the minute (say,
| $4.00/min, based on an average of $240/hr rate for a typical
| consultant from Oracle Consulting) with no expressed or implied
| gurantee to resolve the tech issues (since the techincal problem
| could be OS/HW related). Of course the user will get charged for
| the time they are put on-hold and the time the techical support
| 'researched' the problem. ;-)

I never call Oracle support anymore period even if the site has a support contract. by the time they have figured out a solution or found a person that has some clue what I'm even talking about, I have already solved the problem.

also note that I would never have had any reason to call them if the problems had occured under Linux, because I would have had the tools to pinpoint the problem in a matter of seconds. (the problem I had was with file permissions, but Oracle kept saying that it was out of memory -- which was silly, because the machine had about 120Mb of free memory and had not used a single byte of swap yet).

the people who need to call support, and who will indeed have their problems solved, will do so and they will be able to pay for Oracle's time, but I suspect that many Linux users will do just fine without Oracle, sharing experiences through online forums.

that being said, if Oracle allocated one or two _technical_ people (people who have worked on the Oracle code or who know it intimately) to answer selected postings on a dedicated USENET newsgroup I would suspect that they'd be able to cover 90% of the support the Linux community needs.

hiring two people to answer questions on the USENET costs a minute fraction of what Oracle spends on marketing, but it would buy a *LOT* of goodwill among developers and it would make Oracle a very attractive database for Linux. Oracle probably spends more money on stickers than it would cost to hire an entire platoon of technicians.

do the math. I can't see why they wouldn't jump all over it.

| * I'd doubt Oracle would release source code to the Linux community
| (or any place outside of Oracle, for that matter). I'm sure Sybase,
| Informix, IBM, and a sleuth of other companies would love to look at
| the guts of Oracle.

why would Oracle release the source code? you can buy a source license for Solaris and HP-UX -- that still doesn't mean that Oracle will sell you a source code license for customers running those OSes.

on the other hand; hiring capable Linux people might just lead to an Oracle that is faster and leaner than the current server. quite a few of the people I know that have intimate knowledge of Linux are perfectionists first and peformance freaks second. they write or contribute to code that is ON DISPLAY -- anyone can and will criticize their code.
commercial code is usually not visible to anyone outside the company. this dictates that the code is a product of the programmer-culture that exists within the company. in its state of isolation this culture may produce suboptimal code. not only performance-wise, but also with respect to overall design, cleanliness, maintainability and readability.

perhaps Oracle could do something like what sun did: let Linux-people port Java to Linux.

| BTW, knowing Oracle's architecture and porting strategy, I believe
| it would be fairly simple to port Oracle to Linux.

I have heard the exact same from several sources. I have also heard rumors that Oracle employees successfully _have_ ported Oracle to Linux in order to test it.

| However, unless the business issues I outlined above can be
| addressed, I don't expect to see Oracle on Linux anytime soon...

I think Oracle will have to port their database to Linux sooner or later. Linux is growing and is said to have several times the "market share" of NT. even if only a fraction of the deployed Linux systems are within the business arena it is still probable that Linux has a greater "market share" than NT in the commercial business arena.

I also think that other companies will port their databases to Linux.

personally I would be embarrassed not to port to Linux if I was Larry Ellison. it would only prove that, not only was I out of touch with current trends in the industry, but I would not believe in the message I was trying to put forth about Microsoft.

I have experienced _exactly_ the same situation with WWW. a few years ago, when I tried to convince companies that WWW would be a very good way of publishing information both within the company and globally people told me that it wasn't interesting or that it would never gain the popularity it has today.

some companies even stated at the beginning of the "WWW-era" that they would certainly NEVER waste money on something like that.

guess what happened. companies are basically running amok on the net. it has become a multi-billion dollar industry and *everyone* and everything has a website. companies even have intranets and whatnot.

Bill Gates' attitude towards the Internet was much like the attitude to Linux many software companies have today. he basically screwed up big-time because he didn't understand what he was seeing. Bill Gates _still_ doesn't understand what makes the Internet tick.

in a few years from now we will be laughing at companies who were slow to adopt Linux while talking about crushing Microsoft or delivering open systems.

oh yes, Oracle *will*port to Linux.

-Bjørn Received on Wed May 06 1998 - 00:00:00 CDT