Here is the scenario -
I have an oracle database from which I need to fetch the data through record adapter(SQL pass through). The data is from different tables which are "unrelated" to each other. There are in all 8 such unrelated tables.
Now the approaches that come to my mind -
Approach 1: Create 8 different applications and fetch respective tables. Run ITL over it and push the indexed data to the same Dgraph.
Risk : 1Running multiple applications on the same dgraph at the same time. But this could be mitigated by the processes followed.
Risk : Manageability
Any other risk you see?
Approach 2 :Create 1 application with 8 record adapters and finally join all on the basis of a unique identifier column.
Any risk or challenges do you foresee in this?
If you have any other ideal approach , please share!
Would you just have a switch join? What do you mean by 8 different "applications"
If all the data is just unioned together, you can use CAS (follow the pattern in the product catalog integration module for the deployment template)
If you have more complex joins, alternatively you can do the join in the database which can have a perf benefit over joining in forge.
If you're going to ultimately use 1 MDEX, it doesn't seem like you want to go down the path of 8 different "applications"
Thanks for a quick reply!
Yes, creating 8 applications doesnt seems a good idea and I am more focussing on the second approach -
"Create a pipeline with 8 record adapters that fetch the data from tables and then I will do a switch join and provide it to mapper"
Do you foresee any issues in the above approach?
I am less familiar with how CAS works at the moment. Can you please guide me to the correct document or just brief the approach you suggested.
Thanks a ton for your reply!
The quick start guide that Saleh refers to is definitely the best starting place. Then also check out the doc for the product catalog integration module : http://docs.oracle.com/cd/E35827_01/ToolsAndFrameworks.310/pdf/ProductCatalogDTGuide.pdf