Re: Intercepting Oracle traffic

Why reinvent the wheel?
 The INSTEAD OF triggers can do what you want, I believe, and do not require some odd hacking of the client code..Is there some reason you want to avoid them?

You can also lock down the database except for access through some application that you write and it can control what gets sent completely...

i_sidorin_at_hotmail.com (Igor Sidorin) wrote:

>Thanks! All this actually supports my feelings. But database triggers
>aside, what approach would you recommend? One thought I have is
>plugging in my own module in place of native Oracle client driver
>(that the client application uses), have it do the magic (like
>checking the SQL statements, acting on some, for others invoking the
>real Oracle driver). My concern is the danger of replacing Oracle
>runtime-link module (that would be my only option, as the application
>has been compiled and linked with the compile-time Oracle libraries,
>and I don't have the source code to recompile with my
>driver-substitute). Is the interface to the runtime libraries
>published or is it proprietary (so that only the compile-time
>libraries know how to invoke it)? Has anybody attempted doing this?
>
>Thanks!
>
>Ron Reidy <rereidy_at_indra.com> wrote in message news:<3E1A52F2.1060908_at_indra.com>...
>> Several years ago, I accepted a contract to work at a company trying to
>> this very thing. They said they had "guaranteed" cooperation from
>> Oracle to break the TCP (TNS) stack, and we would be getting "inside"
>> engineering data and help from them.
>>
>> To make a short story (contact length was 4 weeks (originally stated to
>> be 18 months)), the effort was an abismal failure. Furthermore, they
>> are no more (gone, kaput, bankrupt).
>>
>> So, consider using the databases "published" capabilities, not trying to
>> hack the internals to meet some wierd functionality you may need.
>>
>> --
>> Ron Reidy
>> Oracle DBA
>>
>> Igor Sidorin wrote:
>> > smontgomerie_at_hotmail.com (Steve) wrote in message news:<156709aa.0212292106.2592f578_at_posting.google.com>...
>> >
>> >>i_sidorin_at_hotmail.com (Igor Sidorin) wrote in message news:<1b3cae17.0212271212.1e364cc0_at_posting.google.com>...
>> >>
>> >>>Hi,
>> >>>
>> >>>I need to create a proxy for an Oracle database that will look for
>> >>>certain SQL statements and react to them, while passing through the
>> >>>rest of the traffic to the database server. Has anyone done something
>> >>>like this? Is the protocol public? I would appreciate any information.
>> >>>Thanks!
>> >>>
>> >>>Igor
>> >>
>> >>perhaps auditing you mean? maybe look at fine grained access control
>> >>and definitly provide more details if you require meaningful
>> >>assistance
>> >
>> >
>> > Well, here is a scenario. I want to pass through all the traffic
>> > except updates to a certain table. When I catch a statement attempting
>> > to update the table, I may want to send an email somewhere and either
>> > update or not update the table, based on user priviliges, for example.
>> > Or, I may want to look to "Select" from a certain table and return
>> > results from a different database or file - so that it is transparent
>> > to the client. The problem we have is in integrating two products -
>> > one that acts as a client and uses Oracle database (second product).
>> > The first is a black box, so that we can't really do much to it except
>> > posing ourselves as an Oracle database to the client and monitoring
>> > its use of the database. So, in brief, I want to pose myself as a
>> > database so that the client is unaware of my presence, and intercept
>> > certain SQL statements performing some actions based on them. The
>> > actions may be in additions to passing through the traffic (i.e.,
>> > updating Oracle tables based on 'update' SQL statements and then
>> > sending some emails), or intercepting the statement and substituting
>> > the query results from some other data source. Any advice? And no,
>> > using Oracle triggers will not help me (we are trying to stay away
>> > from them).
>> >
>> > Thanks!
>> >
>> > Igor
Received on Tue Jan 07 2003 - 14:11:16 CST