OpenEdge Database Replication Without Using OE Replication o

Posted by v205 on 22-Jan-2019 20:43

Hello. Please forgive is this question sounds a little ridiculous at first, but I am compelled to ask this primarily because of a quote I just received on OpenEdge Advanced Enterprise license for our company. The quote sounded exorbitant in comparison to what we were expecting. 

So, does anyone here have any experience developing a homegrown solution for (considering getting CDC or Fathom/Replication is not an option):

Database Replication (Real-time or near-real-time)

Change Data Capture (perhaps with DB triggers etc.) 

Appreciate your inputs. 

Thanks,

All Replies

Posted by Valeriy Bashkatov on 23-Jan-2019 07:30

Hi,

It all depends on your application, number of users and transactional activity. Also from high availability and reliability requirements.

You can replace OE Replication with AI Replication. But in this case, you will not be able to use the target database as a reporting server, while with OE Replication you can. Also, with AI Replication you will always have a gap between the Source database and the Target database. Because you will need time to archive the AI extents, package them, send them to another server, unpack and run AI Roll-Forward.

OE CDC you can replace with Replication Triggers. But you will have to write separate trigger programs for CUD operations. You can also add your own application tables to save notes about the changes so that the ETL processes can process them later. The main disadvantage here is that in case of high transaction activity in the table your application will run slower than without them. This is because replication triggers are much slower than built into the database CDC triggers.

In any case, OpenEdge is flexible enough to provide you with the features you need, whether it's built-in features or you'll be writing those features yourself. If you have time, you can do everything. All depends on you.

Regards,

Valeriy

Posted by bronco on 23-Jan-2019 08:09

About the replication triggers solution: CUD operations via the SQL engine will *not* be replicated. It's an 4GL-only solution.

Posted by Valeriy Bashkatov on 23-Jan-2019 08:21

Yes, you will have to write Java Triggers for SQL users. This will be "fun"! ;-)

Posted by bronco on 23-Jan-2019 08:27

@v205, Just out of curiousity, how many (db) users are we talking about?

Posted by v205 on 23-Jan-2019 15:48

@bronco We have about 60 concurrent users for one of the DBs. Another DB is also transactional and is purely for the Web site orders. The user count on that is of course, very dynamic...I would say ranging from an average of 10 to 200 at any given point in time.

Thank you everyone for your suggestions!

This thread is closed