Can I use notes in the Professional Data Engineer exam?

No. It is proctored, online or at a test centre. Use this for final revision before exam day only.

PDE Cheat Sheet: BigQuery, Dataflow, Pub/Sub & Storage Choices

A final-revision summary for the Professional Data Engineer exam. Study aid only - no notes in the proctored exam.

The five domains (official weights)

#	Domain	Weight
1	Designing data processing systems	22%
2	Ingesting and processing the data	25%
3	Storing the data	20%
4	Preparing and using data for analysis	15%
5	Maintaining and automating data workloads	18%

The data flow (left to right)

Ingest (Pub/Sub, batch loads) -> Process (Dataflow / Dataproc) -> Store (BigQuery, Bigtable, Cloud Storage) -> Analyse & govern (Dataplex, BI) -> Operate (Composer, Monitoring).

Pick the storage service

Need	Use
SQL analytics over large datasets	BigQuery
High-throughput, low-latency key-based NoSQL (time-series, IoT)	Bigtable
Objects, files, data-lake landing zone	Cloud Storage
Transactional relational app DB	Cloud SQL
Global, horizontally scaling relational DB	Spanner

Pick the processing service

Need	Use
Serverless batch or streaming pipelines	Dataflow (Apache Beam)
Existing Hadoop/Spark or open-source jobs	Dataproc
Ingest and decouple event streams	Pub/Sub
SQL-based ELT transformations in BigQuery	Dataform
Orchestrate and schedule pipelines	Cloud Composer (Airflow)

BigQuery cost and performance

Lever	Idea
Partitioning	Prune by date/range so queries scan fewer rows
Clustering	Sort within partitions to cut bytes scanned
Bytes scanned	Drives on-demand cost - reduce it, reduce the bill
Slots / reservations	Capacity-based pricing as an alternative to on-demand
Materialised view	Precomputed result for frequent, repeated queries

Streaming vs batch

Term	Idea
Streaming	Continuous processing as events arrive (Pub/Sub + Dataflow)
Batch	Scheduled bulk loads and transforms
Windowing	Grouping streaming data into time windows for aggregation

Governance and operations

Term	Idea
IAM	Roles and least-privilege access for users and service accounts
Dataplex	Organise, govern and discover data across lakes and warehouses
Cloud Monitoring / Logging	Metrics, alerts and logs to operate pipelines
Service account	Identity pipelines use to authenticate to GCP

Exam facts at a glance

Item	Value
Duration	120 minutes (2 hours)
Questions	40-50 (per the official exam guide)
Passing score	Not published by Google (pass/fail)
Format	Multiple choice and multiple select; online-proctored or test centre
Fee	US$200 + tax (recertification US$100 - confirm)
Validity	2 years

BigQuery vs Bigtable - the one to get right

BigQuery is the serverless warehouse for SQL analytics on large datasets; Bigtable is wide-column NoSQL for high-throughput, low-latency key-based access such as time-series or IoT. The exam tests when to use each, not depth in one. If the scenario is “analyse with SQL,” think BigQuery; if it is “millions of reads/writes by key with low latency,” think Bigtable.

Google Cloud Professional Data Engineer (PDE) Cheat Sheet