When/how should I swap batch jobs for independent jobs?

I'm wanting to send simulations to a remote cluster with 32 workers. However, I'm running into problems sending batch jobs in that when I retrieve my data, I can't bring in more than 2GB of data. I've been trying to come up with a way to send individual jobs to each worker and retrieve each worker's data separately. Here's an example of what I've been doing:

% set up variables

 for parameters

 batch

Here's what I'm thinking about doing:

% set up variables

 parfor the_simulations

 createJob

 createTask main_sim_data

 createTask secondary_data_based_on_sim_data

 submit job

Is this an appropriate change, to go from a batch job to independent jobs using the createJob & createTask parallel functions? When does it make more sense to do batch instead of independent jobs, and vice versa?