GCS: Object storage. Similar to S3. Stores raw and processed data.
BigQuery: Serverless data warehouse. No clusters to manage. Pay for queries and storage separately. Nested/repeated fields handle semi-structured data well.
Dataflow: Managed Apache Beam. Batch and streaming in one API. Auto-scales based on workload.
Dataproc: Managed Spark/Hadoop. Use for Spark workloads.
Pub/Sub: Fully managed messaging for event ingestion. Handles scaling automatically. Different ordering and retention semantics than Kafka.
GCP shines for analytics workloads. BigQuery's serverless model simplifies operations.