[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: carp + pfsync + sqlnet (Oracle)

[email protected] said:
> On Mon, Jun 21, 2004 at 02:28:26AM -0500, James Cammarata wrote:
>> At 06:00 AM 6/16/2004, [email protected] wrote:
>> >Does anybody protect any oracle rdbms (sqlnet protocol) using
>> >obsd 3.5 + carp + pfsync ? Does it work ? Is it problematic ?
>> I assume you want to do a redundant DB correct?  Databases are not
>> suited
>> to this kind of failover, due to the lack of consistency between the
>> files
>> on different disks.  Your best bet is to use Oracle's built in
>> redundancy
>> (as expensive as that may be).  Creating a stand-by server is not cheap,
>> but that is the kind of redundancy you want.
> I don't want to use stand-by server. I am aware of it and what it does.
> Let assume I have oracle rdbms behind obsd firewall (working as a bridge)
> and
> it works ok. Now obsd is a single point of failure (SPOF) and if it
> crashes/hangs (due to e.g. hardware failure) nobody can access oracle.
> If I setup second obsd fw in a pair with the first (using carp + pfsync)
> the obsd is not a SPOF but I am not sure if such configuration
> influences sqlnet traffic somehow.
> przemol
Not that I write the following from memory as it was a while ago. So you
maybe have to double check me on this.
If you just talk about single sessioned TCP traffic it's ok I think. But I
think Oracle SQL*Net when running the DB listener on a windows server can
be set up in two ways (earlier just one). The ugly way is like active ftp
with a new connection going back for data (you can see the problem with
this "mode"...). With Oracle on unix as far as I know this ugly thing does
not exist. But nowdays I think you can choose "mode" on windows as well.
I have earlier set up SQL*Net traffic through a "stunnel" encryped tunnel
and therefore checked this up.
So if you run the SQL*Net traffic as single sessioned TCP traffic at least
I can't see a problem...
Hope this info can be of any help.