Re: OCI libraries for Linux

[twod_at_not.valid]
|
| Mr Bjorn Borud (borud_at_itea.unit.no) wrote:
| : this is becoming more and more of a farse. at least for Sybase you
| : can get the client libraries separately and for free. at least for
| : Linux -- I do not know if the client libraries are available
| : separately for the other platforms.
|
| It may be a farce, but what you are trying to do is use a not exactly
| straight-forward, so you have to take the rough with the smooth.

well, it should be straight forward. if I wish to create software for 4 different platforms and the software needs OCI in order to communicate with the database running on one platform, why should I be forced to buy an Oracle server for all four platforms? of course I won't.

| : price. (buy a "single seat" server license for about $200,- and you
| : have the libraries). of course, you can run them on Linux as well as
| : other operating systems.
|
| For $200 you can also get a commercial license for 'mSQL'. Amongst
| the many databases I have installed, I use a beta version of mSQL
| v2.0 with the DB[ID] modules of PERL5. It performs quite well,
| actually very well for smallish amounts of data - probably better
| than Oracle. It may be worth a peek - www.Hughes.com.au - but I
| don't think that it's a scaleable as Oracle or other full-monty
| commercial databases. It's available as 'C' source code, so is
| highly portbale.

I am familiar with mSQL and I have used mSQL 1.0.16 for applications where light database usage has been involved. however the subset of SQL implemented in both 1.0.16 and the 2.0 betas of mSQL is a far cry from that of Solid, which even sports a procedural language, sequences, ODBC and of course a more complete set of SQL commands.

check out http://www.solidtech.com/

| : | Knowing the iBCS emulation under Linux, perhaps see if the SCO
| : | Workgroup Server 90 day trial software is shipped with them.
 

| : this sounds sort of "half way legal" and I am not sure I wish to risk
| : doing something that might be illegal. thanks for the tip though.
|
| Depends on the licensing agreement that you have with Oracle. I work for a
| company with a license for all of Oracle's products, we can use them as we
| see fit, but if we tried to get support for the SCO executables on
| Linux ...

have you informed Oracle of your wish to have Oracle on Linux? please do so.

| : this is very frustrating to me since I initially had a very positive
| : attitude about using Oracle.
|
| Errm, so you assumed their was a linux port, or linux was your OS of
| choice ?

no I started work on OCI under AIX then HP-UX well aware that Oracle didn't support Linux. what frustrates me is the uncleanliness, the mess and the pain of developing OCI applications since I am not used to erroneous error messages (to that degree anyway) and various fiddling around with the environment that is totally uncalled for.

also it seems that file permissions and such, being rather important, is not well defined and that there is no recipe as to where one might look should trouble arise. the documentation is rather sparse in some aspects, with a profound allergy to mentioning a single piece of information in what might be considered "the wrong spot" -- although such a mortal sin might be useful to the reader who doesn't feel like carrying around the full manual set for Oracle at all times... :-)

| : | Going elsewhere is an option and will hit Oracle where it hurts, in the
| : | year-end $$ totals.
 

| : really, because I am under the impression that there are more
| : well-trained DBAs using DB2 and Oracle then there are DBAs experienced
| : in using Sybase, Adabas or even Solid.
|
| Ever tried to find a well-trained Oracle DBA ?
|
| We get people at interviews claiming 3 years DBA experience who cannot
| explain what a rowid is ! [I'd like to say that this is a joke, but sadly
| it is not]

well, I have to admit that I have never been in the position where I needed to hire a DBA, so I wouldn't know about the DBA job-market. especially not that in the US. our customers usually have their own "home grown" DBAs since the banking and insurance business employs quite a few database people and thus creates an environment where apprenticeship is probably the way new DBAs are broken in. it seems to work, although an amazingly large number of DBAs are unfamiliar with modern computer technology having talked to their databases on mainframes since the dawn of time.

| : I'm on the doorstep of choosing database technology for 3 or 4 future
| : projects right now. projects where the database will be very
| : important because it will be the backbone of the system. I am willing
| : to trade functionality for ease of use, stability and portability.
 

| : I really don't know if Oracle is what I should be using. since at
| : least two of the projects involve several thousand client machines
| : of various kinds this suggests I won't be using Oracle -- at least not
| : directly without some sane middleware.
|
| Take a deep breath.
|
| Take a step back.
|
| Take another step back.
|
| Have a think what you are trying to achieve.
|
| OK, let that breath out.
|
| I would suggest that you and your system architects get a clean whiteboard
| and a few coloured pens and have a think about what you are trying to
| achieve.
|
| Phrases such as 'the backbone of the system' and ' portability', do not
| spring to mind when you mention 'Linux' and 'Oracle' in the same sentence.
| Sure Linux is stable and System V'ish; sure, Oracle is stable and portable,
| but the match was not made in heaven.

actually, the database itself would most likely not run on a Linux system, but the "nodes" might well be running Linux. also, we would prefer being able to use Linux for development since it is in most cases easier to debug software under Linux. also, in my experience, software written on Linux is usually easier to port to other UNIX platforms, and portability is of importance.

today I find myself writing part of the software under Linux, postponing the parts that need OCI, and then moving development to whatever platform we have OCI for. I shouldn't have to do this.

| From what I can gather from your previous posts, you will have an
| Oracle RDBMS sitting somewhere on a server that will be accessed
| from many remote sites/operating systems using OCI and, presumably,
| SQL*Net ?

something along those lines, yes.

| I'm prepared to go off-line with this, as it could get tedious for
| others. but how about using Oraperl ? It's as portable as PERL; in
| my experience offers the same performance as 'averagely written'
| Pro*C/OCI (ie no host arrays, although that is being worked on,
| albeit it slowly); allows for great portability - platform and
| database; allows development work with minimal effort (ie
| compile/port the DBD modules for a given databse/platform once); is
| freely available; is supported in the usenet tradtion of mailing
| lists that get right to the developers.

the problem is that the software that talks to the database is integrated with software that is doing something else as well. adding another layer of tape and glue in a different language would make me somewhat uncomfortable -- although this is what Oracle would force us to do in any case. in C. which is why Oracle is becoming a less attractive solution the more I look at it.

| When starting up a project, you should ascertain what are the
| critical components that have to provide the best performance and/or
| the most flexibility. It is these critical components (ie the
| database engine) rather then the generic components (ie OS's) that
| should form the key part of your solution.

as for the "nodes" there are certain aspects of Linux that makes Linux more attractive than the other options and it has nothing to do with Linux being free. I'm sorry that I can't elaborate further on that specific aspect, since it would clarify my wish to run Linux on the "nodes", but the OS is in some respects just as important as the existance of a RDBMS. note "a RDBMS" since our database needs can be met with other databases than Oracle's.

| Naturally, the generic components should match/enhance the critical
| components, but you have to make sure that the critical components
| are able to give their best performance. I somehow feel, that things
| are the other way around in your current plan.

in my current plan (or rather "our") the situation is that Linux is attractive to run on the nodes and during the development of other components in the system. partly because it is an environment well suited for development, partly because it has better performance.

if we are unable to find a RDBMS that all nodes can talk to directly we have to seek out a solution that accomplishes this indirectly. if none exists we have to create one -- at increased cost.

| As much as I admire Linux/FreeBSD/NetBSD/Coherent/ foo, it is a sad
| fact of (commercial) life that for certain applications they are,
| through no fault of their own, not yet read to go into full-monty
| production in some areas.

I think this has to do with the prejudice with which Linux is often met. I have seen much talk about Linux not being ready for commercial use, and I have to say that I am frustrated when hearing the same old talk over and over. a lot has happened to Linux the last year. I wish these so called analysts woke up and grasped the fact that Linux is evolving several times faster than everything they are used to. and it doesn't mean that things next year won't be backward compatible like in the Microsoft industry.

unofficial reports say that programmers at Oracle have done some testing on Linux and found that it works well on Linux. well? why don't they release Oracle for Linux and give the product a couple of years? they tried to make a browser? didn't they? everyone could have told them that it would never fly. they also made a web-server. what return did they have on that? from what I can see in the Netcraft statisticts it's not a very popular product. why are they so willing to spend money on that instead of appeasing the professional Linux users?

professional Linux usage is mainly "invisible" today since it is largely used for "invisible" tasks.

I'm not a Linux zealot, but it does freak me out that companies like Oracle are silly enough to ignore Linux simply because their analysts can't understand it.

-Bjørn

-- 
 Bjørn Borud <borud_at_pvv.org>      | "The Net interprets censorship 
 <URL:http://www.pvv.org/~borud/> | as damage and routes around it."
 UNIX person, one of "them"       |         - John Gilmore