gcloud ml-engine local train - run a Cloud ML Engine training job locally This command runs the specified module in an environment similar to that of a live Cloud ML Engine Training Job.
This is especially useful in the case of testing distributed models, as it allows you to validate that you are properly interacting with the Cloud ML Engine cluster configuration.
Reference: https://cloud.google.com/sdk/gcloud/reference/ml-engine/local/train

To centralize unstructured data from various sources into Cloud Storage using a GUI-based solution while allowing the use of your own encryption keys, Cloud Data Fusion is the most suitable option. Here's why:
Cloud Data Fusion:
Cloud Data Fusionis a fully managed, cloud-native data integration service that helps in building and managing ETL pipelines with a visual interface.
It supports a wide range of data sources and formats, including Apache Parquet and CSV, and provides a user- friendly GUI for pipeline creation and management.
Custom Encryption Keys:
Cloud Data Fusion allows the use of customer-managed encryption keys (CMEK) for data encryption, ensuring that your data is securely stored according to your encryption policies.
Centralizing Data:
Cloud Data Fusion simplifies the process of moving data from on-premises and cloud sources into Cloud Storage, providing a centralized repository for your unstructured data.
Steps to Implement:
Set Up Cloud Data Fusion:
Deploy a Cloud Data Fusion instance and configure it to connect to your various data sources.
Create ETL Pipelines:
Use the GUI to create data pipelines that extract data from your sources and load it into Cloud Storage.
Configure the pipelines to use your custom encryption keys.
Run and Monitor Pipelines:
Execute the pipelines and monitor their performance and data movement through the Cloud Data Fusion dashboard.
Reference Links:
Cloud Data Fusion Documentation
Using Customer-Managed Encryption Keys (CMEK)

Cloud Dataproc provides out-of-the box and end-to-end support for many of the most popular job types, including Spark, Spark SQL, PySpark, MapReduce, Hive, and Pig jobs.
Reference: https://cloud.google.com/dataproc/docs/resources/faq#what_type_of_jobs_can_i_run

In a Cloud Bigtable architecture all client requests go through a front-end server before they are sent to a Cloud Bigtable node.
The nodes are organized into a Cloud Bigtable cluster, which belongs to a Cloud Bigtable instance, which is a container for the cluster. Each node in the cluster handles a subset of the requests to the cluster.
When additional nodes are added to a cluster, you can increase the number of simultaneous requests that the cluster can handle, as well as the maximum throughput for the entire cluster.
Reference: https://cloud.google.com/bigtable/docs/overview