I have also a spark cluster created with google cloud dataproc. Source: about 1 year ago

Specifically, we heavily rely on managed services from our cloud provider, Google Cloud Platform (GCP), for hosting our data in managed databases like BigTable and Spanner. For data transformations, we initially heavily relied on DataProc - a managed service from Google to manage a Spark cluster. - Source: dev.to / about 2 years ago

With that, the best way to maximize processing and minimize time is to use Dataflow or Dataproc depending on your needs. These systems are highly parallel and clustered, which allows for much larger processing pipelines that execute quickly. Source: over 2 years ago

The Apache Spark / Databricks community prefers Apache parquet or Linux Fundation's delta.io over json. Source: 6 months ago

Apache Parquet (Parquet for short), which nowadays is an industry standard to store columnar data on disk. It compress the data with high efficiency and provides fast read and write speeds. As written in the Arrow documentation, "Arrow is an ideal in-memory transport layer for data that is being read or written with Parquet files". - Source: dev.to / about 1 year ago

Googling that suggests this page: https://parquet.apache.org/. Source: about 1 year ago

You should also consider distribution of data because in a company that has machine learning workflows, the same data may need to go through different workflows using different technologies and stored in something other than a data warehouse, e.g. Feature engineering in Spark and loaded/stored in binary format such as Parquet in a data lake/object store. Source: over 1 year ago

This section will teach you how to read and write data to and from a variety of file types, including CSV, Excel, SQL, HTML, Parquet, JSON etc. You’ll also learn how to manipulate data from other sources, such as databases and web sites. Source: over 1 year ago