A namespace contains all database-related components:
In our case, the “manhattan-redshift-namespace” contains the “dev” database.
A workgroup is the compute part for running queries:
The “manhattan-redshift-workgroup” has 4 base RPU and an endpoint for Query Editor v2 connection.
They are separated to manage storage and compute independently.
RPU (Redshift Processing Unit) measures the compute power of Redshift Serverless, including CPU, memory, network, and query processing power.
We chose 4 RPU because we’re only testing queries on taxi data (~a few hundred MB).
Redshift Serverless charges based on RPU-hours used. Higher capacity means higher costs for heavy queries. For hands-on work, 4 RPU helps save money.
Redshift needs an IAM role to:
The role is created with AmazonS3ReadOnlyAccess and AWSGlueServiceRole permissions to read S3 and Glue.
Redshift Serverless has a $300 free trial credit for 90 days for new accounts. This is separate from the AWS Free Tier $200.
We use this credit to avoid charges during learning.