Oracle Versions Communication

Question

Hi Guys,
this is my first post, so apologies in advance if I've selected the wrong part of the forum. . .
We are currently undergoing a project to move the database for one of our systems from 8i on unix over to 10g on windows 2k3 (we no longer have any unix support since ms is the company standard - a decision far above our heads that we cannot question). We have had a few problems with past attempts, but these may be explainable by outside factors (i.e. we discovered that some ports were blocked around the same time we made one attempt, and there was a virus outbreak on another go - like I say, the company standard is ms), so people are questioning whether the 'new technology' is compatible, or if we should try using older versions of OSes & oracle.
As I understand the theory, the operating system & oracle version should not cause any issues to our system if only the server side is changed, and the only change made to clients is to point them to the new server using their tns names file. Since all communication is via tcp/ip, and the database before and after is oracle, I would expect that if the same messages are being sent, the messages received should be the same, though the black-box server processing side of things may be different - i.e. aside from knowing about a change of server name and faster response, our clients should not be able to tell that there has been an upgrade, even when doing a binary level comparison of the responses. Does anyone know of any difference that could exist in the replies (i.e. would the later version of oracle maybe try to use a different data structure, or does the reply include some sort of tag to indicate what version of oracle was run on the server)?
Hope that all makes sense. . .
Many thanks in advance for your help,
JB
(Hope the bold helps to separate the question from the background padding ;) )

Answer 1

Hi Arulselvaraj,
That's kind of what I'm asking, but at a more technical level. I am not worried about the actual user's, but the client software (in this case a very old and heavily customised version of JDE OneWorld). I am wondering if the binary data returned to a client by a server running 8i on Unix would look exactly the same as the binary data returned by a server running 10g on windows 2k3 when both servers have the same data and are sent the same query by the same client? In other words, would a migration from one to the other be transparent to the client application?
Hopefully that makes a bit more sense; not the easiest of issues to explain.
Thanks again,
JB