Re: SQL*Net and Redundant Networks

In article <3onabr$qaa_at_zaffiro.sasib.it> sergio_at_zaffiro.sasib.it (Sergio Bertolini) writes:
>From: sergio_at_zaffiro.sasib.it (Sergio Bertolini)
>Subject: SQL*Net and Redundant Networks
>Date: 9 May 1995 09:50:03 +0100

>We have a redundant configuration like this :
>
> Net 128.1.1.0
> ------------------------------------------------
> | |
> |128.1.1.1 |128.1.1.2
> [Server] [Client]
> |128.1.2.1 |128.1.2.2
> | |
> ------------------------------------------------
> Net 128.1.2.0
>
 

>i.e a server and a client machine which both have 2 ethernet board facing two
>different TCP/IP networks ( for redundancy ).
 

>On the server we have the Oracle Server (7.0) with its SQL*Net V2 listener and
>we would like it could be accessible by the client machine, which has just
>clients compiled with libsqlnet, from both sides, or better , when a network is
>down we would like the server still be reached by the other side.
 

>Any hint ?
 

>Thanks in advance.

I know you'd like this to be automatic and I don't think you'll ever get that at the Oracle level. That belongs at the network level. However you can run two listeners, one for each address. Then code to try both in your app.

Connect listner1
If failed
  Connect listner2
end if

You could also put error trapping code in your app and if the connection is lost retry the other address. Or just let the user restart the app and it will reconnect to the other link.

If you want to get really slick you could figure out a way for the client to find out how many connections are made through each side and if the other side has less, move. You could use a table that they 'log' into. Of course you'd have the failed connections to cleanup somehow.

Like I said this functionality belongs at the network level, not the application. Good luck, Lawrence... Received on Wed May 17 1995 - 00:00:00 CEST