Batching of large volume of journal entries for import into Cloud GL from external system

Summary: We will be building an integration to send a large volume of journal entries from a custom financial reporting system for import into the Cloud GL. I believe we are looking at approx 500,000 lines total, and trying to determine the best way to split these up to send to Cloud GL.

Content (required): A few options I can think of are:

1) Create a GL batch per retail location (with one or a couple journal headers within the batch, and somewhere around 2-200 lines per journal) - this would result in approximately 4000 batches.

2) Group entries at a much higher (TBD) level, resulting in many fewer batches and then create a journal header per retail location.