Re: Replication or Database Link ? (Web-Architecture with EJB)

Folks,

thanks for all your replies.
To bring more detail into the discussion and add one more aspect: Our WEBDB is in fact a database that will be accessed by internet users. Security-wise both databases are behind a firewall and the WEBDB is accessed from an application server via IIOP. That means that our WEBDB is an Oracle8i with EJB-logic running inside the database and the CALLDB is an Oracle8.0.5 both under Solaris.

Now, we provide additionally some stored procedures that work on our MEMBER table and need to be accessed from some other PL/SQL stored procedures as well as from the EJB. Using a database link, that would mean that I need to call a stored procedure on another database via a database link. How ?

I see the following solutions - neither one seems to be very attractive:

1. Duplicate the stored procedures and replace e.g. "select * from member" by "select * from member_at_CALLDB".
2. Let the EJB make an additional connection to the CALLDB and execute the stored procedure on CALLDB.

I believe that the latter one seems to be more appropriate ?

Any comments or ideas ?

Thanks again, Timo.

In article <EjBs3.152$5W.3156230_at_news1.van.metronet.ca>,   "DET" <NotValidBecauseOfSpammers_at_ipipeline.net> wrote:
> First problem:
>
> With synchronous replication, if one database goes down, you can't do
> updates to the other until you either bring up the downed database or
turn
> off the synchronous replication -- which brings up synchronizing
problems
> when you finally fix the crashed database.
>
> I assume one database is on the internet and one is on your internal
LAN.
> How are you getting the connection across the firewall? what security
are
> you implementing?
>
> tsalzsie_at_my-deja.com wrote in message <7ots95$jes$1_at_nnrp1.deja.com>...
> >Pete,
> >
> >thanks for your reply. There are some semantic as well as security
> >reasons that we need two databases. The databases are in the same
> >network (in other words in the same data center) and afaik are
connected
> >via a 100 Mbit network.
> >One of the reason some people in our department tend to use
synchronous
> >replication is availability - in other words, WEBDB is still availabe
> >when CALLDB crashes and vice versa. However, I'm not sure how to
handle
> >these large batch jobs at night - how replication reacts or if it
makes
> >sense to disable replication while these batch jobs are running.
> >So, database links seems to be the most appropriate solution with
> >respect to simplicity, reliability and performance.
> >
> >Any other recommendations or things we should look for ?
> >
>
>

Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't. Received on Wed Aug 18 1999 - 09:15:47 CDT