GCP Data Engineer Pro study plan

Compare batch versus streaming approaches and select an appropriate processing service for three written business scenarios

Build a one-page matrix mapping each GCP data service (Pub/Sub, Dataflow, Dataproc, BigQuery, Bigtable, Cloud Storage, Composer) to the design constraints it satisfies

Draft a partitioning and file-layout strategy for a 10 TB-per-day event dataset, including schema and late-data assumptions

Estimate cost, latency, and SLA tradeoffs between BigQuery, Bigtable, and Cloud Storage for the same read/write workload

Separate a design into a streaming ingestion path (Pub/Sub) and a batch ingestion path (Storage Transfer Service or Datastream), documenting the handoff

Build and deploy a working Pub/Sub to Dataflow to BigQuery streaming pipeline in a sandbox project

Make the pipeline idempotent and prove it by replaying duplicate messages

Configure dead-letter topic handling and retry policy on a failing subscription

Define schema-evolution rules for the BigQuery destination table

Implement fixed, sliding, and session windows in Apache Beam and compare emitted results on the same late-arriving dataset

Add watermarks, allowed lateness, and triggers to a pipeline and record how each setting changes output

Tune a Dataflow job's worker count, disk size, and shuffle behavior, then record the effect on runtime and cost

Write unit tests for Beam DoFns using TestPipeline that assert correct handling of late and duplicate records

Run a Dataproc Spark job that reads from Cloud Storage, transforms data, and writes to BigQuery

Choose between Dataproc, Dataflow, and Dataproc Serverless for three workload descriptions and justify each choice

Model a star schema in BigQuery using partitioning, clustering, and nested/repeated fields

Load the same dataset with batch loads and streaming inserts and measure the cost and latency difference

Partition and cluster a BigQuery table and measure bytes scanned before and after on the same query

Implement a Cloud Storage lifecycle policy with storage-class transitions and verify objects move as expected

Design a Bigtable row key for a time-series access pattern and explain the hotspot risk it avoids

Compare Bigtable, Spanner, Firestore, and BigQuery for three distinct access patterns

Enforce column-level access control using policy tags on a BigQuery table

Write advanced BigQuery SQL using window functions, array and struct manipulation, QUALIFY, and approximate aggregation

Build a materialized view and enable BI Engine, then measure the query-time improvement

Apply a data-quality check with Dataplex or SQL assertions and fail a pipeline when it is violated

Compare a denormalized single-table model against a normalized join for the same analytical query

Orchestrate a three-task pipeline in Cloud Composer with explicit retries and an SLA alert

Create Cloud Logging metrics and alerting policies for pipeline failure and data-freshness thresholds

Automate deployment of the pipeline with Cloud Build and Terraform

Configure least-privilege IAM service accounts for each pipeline component

Set a budget alert and document when to use slot reservations versus on-demand pricing

Re-run the four weakest labs from memory under a time limit

Close the five lowest-scoring objectives from the official exam guide, one per study block

Complete two full-length timed mock exams and log every error by domain

Rewrite a personal cheat sheet containing only the 30 facts you repeatedly miss