Re: Mosaic <-> Oracle

beyret_at_chbs.ciba.com (Tim Eyres) writes:
> Hi y'all
>
> I know some people out there have created interfaces
> between Mosaic and Oracle databses (which is my
> current project) so could anyone give me some
> assistance on the matter, ie: what's the best way
> of doing it ? Code people have written would
> be most appreciated.
>
> Tim (beyret_at_chbs.ciba.com)
>

Hi,

I've had a lot of replies to the above mail asking for information so I figured that I may as well start a little thread going.

This an internal feasability study I made for Ciba-Geigy so I figured that it would be usefull to start this off. Any comments you may have on this would be much appreciated

> Tim Eyres 7/11/94
>
> FORUM <-> WWW gateway
>
> Feasability Study
>
> This report discusses the possiblities of constructing a gateway between the
> FORUM conference database and the SCS WWW server. It covers the initial ideas,
> the subsequent investigation, the conclusions and the possibly courses of action
> to be taken.
>
> Inital Ideas
>
> The problem is to create a Mosaic interface to an Oracle database which uses
> SQL as the intermediate language. The main difficulty is how the data would
> be transfered from the Mosaic client to the Oracle database. There are three
> options available,
>
> 1 Create a program/script that connects directly to the Oracle
> server and generates a file of SQL requests which when
> processed by SQLPLUS, returns the ouput from the
> Oracle server as an ASCII file.
>
> 2 Create a script which would generate a file of SQL statements
> which could be sent to the FIZ machine ( Machine which also uses
> the same database ) and processed by a server process running on
> that machine.
>
> 3 Create a script which would generate a file of SQL statements
> which could be sent to the Oracle server and processed on that
> machine.
>
> 4 Using Pro * C, connect directly to the Oracle server and execute
> embedded SQL on the Oracle server.
>
> In all of these cases we have to process HTML into SQL (From Mosaic to Oracle),
> and then convert the ASCII into HTML (From Oracle to Mosaic). These problems
> are the same in all instances so the only decision to be made is how data should
> be transfered between the httpd and Oracle server.
>
> The following diagrams show the logical and physical structure which these
> three ideas created.
>
> Idea 1
>
> There should be a picture in here but because of my naff editor I can't output
> in postsript hence this will have to do. Hopefully you can make sense of it
> all anyway
>
> Further Investigations
>
> For this we have no code so we would have to construct our own. In addition,
> we need to install SQL*PLUS and SQL*Net (for connection to the database) on
> the compbio machine ( machine running the httpd process)
> so that we may process the SQL statements via the Oracle
> server. This would cost money and time to install the software.
>
> We do not, however, have to use the FIZ machine which will save us run-time.
> In addition an ASCII file for the SQL statements can be easily changed, and
> SQL*PLUS and SQL*NET will allow us to connect to other Oracle databases.
>
> Idea 2
>
> Again another picture...
>
> Further Investigations
>
> For this we have no code so this we would have to construct our own. We also
> need to have a program on the FIZ machine which will receive the SQL request
> file, process it via the Oracle Server, and return the ASCII result file. Whether
> such a procedure is already available I do not know, but it could result in
> a slow system compared to directly linking with the Oracle Server. We may also
> need to pay FIZ for runtime costs.
>
> We do not, however, have to install any software to CompBio or worry about
> the location of the Oracle server and it's
> databases.
>
> Idea 3
>
> And again...
>
>
> Further Investigations
>
> For this we have no code so we would have to construct our own. We also need
> to have a program on the Oracle machine which will recieve the SQL request
> file, process it and return the ASCII result file. Whether such a procedure
> is already available I do not know, and using this method we have to keep track
> of the location of the FIZ database. Could there possibly be security consideratoins
> with regard to us connecting to the Oracle server ?
>
> We do not, however, have to install any software to CompBio.
>
> Idea 4
>
> And again...
>
> Further Investigations
>
> For this we have code available hence saving time. However, to use the embedded
> SQL we require a copy of Pro * C to compile the code. We also need SQL*Net
> to connect to the Oracle database. With SQL*Net and Pro * C we will then have
> the ability to connect to other Oracle databases. This also allows the direct
> display of results with no intermediate ASCII files required.
>
> We do not, however, have to use the FIZ machine which will save us run-time.
>

See ya !

Tim

(beyret_at_chbs.ciba.com) Received on Mon Nov 07 1994 - 09:35:17 CET