Understanding Spark job submission

[Muhammad Gelbana]

I'm wondering how does Spark adapter work ?
Where exactly does it communicate with Spark ?
Is it targeting embedded Spark instances only or can use Spark clusters ?

Thanks,
Gelbana