Five Stars

Question on variable scope: subjobs, joblets, tRunJob, tJavaRow, tJava

I would like some information on variable scope when working within a subjob, working across multiple subjobs in the same job, when executing a joblet vs a tRunJob.  
How are variables shared, context shared and are globalMaps shared or not? 
I am also interested in how these are impacted when you parallelize a flow, for example using round robin allocation to threads from an input component.  
I am trying to create a complex, multi level mongoDB document, and want to speed it up through parallelization, but I have a high dependency on storing documents I am creating in the globalMap, so I can carry on down the flow, execute more SQL parameterised by globalMap variables and build an embedded document.
It works, but it is I want to make the first tOracleInput parallelized, so each row goes out to a different thread....but then my globalMaps will be trampling all over each other.
I tried writing things into the globalMap using a combination of the threadID and the object key name, using globalMap.get("tCollector1_THREAD_ID"), but I don't understand how this gets populated...surely if its in the globalMap, the threads are trampling all over this value all the time also?
Perhaps I should create joblets or subjobs to make sure I have separate scopes, and pass things by reference using contexts? 
Is there any advice or documentation on this please?

Carpe Diem

Re: Question on variable scope: subjobs, joblets, tRunJob, tJavaRow, tJava

Would you mind sharing your current job design screenshot into forum?
Best regards
Don't forget to give kudos when a reply is helpful and click Accept the solution when you think you're good with it.