Google Cloud • PCDE
Google Cloud Certified - Professional Cloud Database Engineer (PCDE) Practice Test
Validates expertise in designing, creating, managing, and troubleshooting Google Cloud databases, with focus on scalable and highly available database solutions spanning multiple database technologies.
Exam Details
Questions
608
Duration
120 minutes
Passing Score
Not publicly disclosed
Difficulty
ProfessionalLast Updated
Jan 2025
Topics Covered
Database DesignCloud SQLCloud SpannerBigtableFirestoreData MigrationDatabase SecurityMonitoring and Troubleshooting
Exam Domain Breakdown
Design innovative, scalable, and highly available cloud database solutions32%
Manage a solution that can span multiple database technologies25%
Migrate data solutions23%
Deploy scalable and highly available databases in Google Cloud20%
Exam Overview
The Google Cloud Certified Professional Cloud Database Engineer (PCDE) certification validates the expertise of database professionals who design, create, manage, and troubleshoot Google Cloud databases. Holders of this credential demonstrate the ability to translate complex business and technical requirements into scalable, cost-effective, and highly available database solutions spanning multiple Google Cloud database technologies — including Cloud SQL, AlloyDB, Cloud Spanner, Cloud Bigtable, Firestore, and Memorystore.
The exam covers the full database lifecycle on Google Cloud: from architecting multi-region, regionally resilient deployments to selecting the right storage paradigm (relational, NoSQL, in-memory, vector) for a given workload. Candidates must demonstrate proficiency in database migration strategies, security and IAM configuration, backup and recovery planning (RTO/RPO/PITR), performance monitoring, and troubleshooting. As of its most recent update, the exam also includes coverage of generative AI use cases with Google Cloud databases.
Who Should Take This Exam
This certification is designed for experienced database professionals and cloud engineers who work hands-on with Google Cloud database services. Typical roles include Database Administrators (DBAs), Database Engineers, Cloud Architects, and Backend Engineers who are responsible for provisioning, managing, and optimizing database infrastructure on Google Cloud.
Google recommends candidates have at least 5 years of overall database and IT experience, including a minimum of 2 years of hands-on experience working specifically with Google Cloud database solutions. It is best suited for professionals who regularly make architectural decisions involving trade-offs between relational and NoSQL systems, manage migrations from on-premises or other cloud platforms, and are responsible for database availability and disaster recovery.
Prerequisites
There are no formal prerequisites required to register for the exam. However, Google strongly recommends candidates possess 5 or more years of general database and IT experience, with at least 2 years of practical, hands-on experience working with Google Cloud database products such as Cloud SQL, Cloud Spanner, Bigtable, and Firestore.
Candidates should be comfortable with SQL and NoSQL concepts, IAM and database security models, replication and high availability configurations, data migration tooling, and performance tuning across multiple database engines. Familiarity with the Google Cloud Console, gcloud CLI, and Cloud Monitoring is also beneficial before attempting the exam.
Exam Format
The Professional Cloud Database Engineer exam consists of 50–60 multiple-choice and multiple-select questions. The exam must be completed within 2 hours (120 minutes). It is available in English and can be taken either online via remote proctoring or in person at an authorized Kryterion testing center worldwide. The registration fee is $200 USD plus applicable taxes.
Google does not publicly disclose the passing score threshold. The exam is scored holistically across all domains, so candidates are not required to achieve a minimum score in any individual domain — balanced preparation across all four domains is recommended. Certification is valid for a defined period, after which candidates may renew within an eligible renewal window.
Skills Measured
- 1.Domain 1 – Design innovative, scalable, and highly available cloud database solutions (~32%): Covers selecting the appropriate database product (Cloud SQL, AlloyDB, Spanner, Bigtable, Firestore) based on business and technical requirements; sizing compute and storage for performance and cost; evaluating trade-offs between SQL, NoSQL, and vector storage; designing multi-region, regional, and zonal deployment strategies; planning for high availability and disaster recovery; and addressing generative AI use cases with Google Cloud databases.
- 2.Domain 2 – Manage a solution that can span multiple database technologies (~25%): Covers implementing IAM roles and database-level access controls; monitoring database performance using Cloud Monitoring and Query Insights; managing backup and restore operations including point-in-time recovery (PITR); defining and meeting RTO/RPO objectives; automating database operations; and troubleshooting performance and connectivity issues across Cloud SQL, Spanner, Bigtable, and Firestore.
- 3.Domain 3 – Migrate data solutions (~23%): Covers planning and executing database migrations with minimal downtime; selecting appropriate migration tools (Database Migration Service, Datastream, pg_dump, etc.); performing schema and query conversion including DDL/DML transformation; developing fallback and rollback plans; and handling heterogeneous migrations between database engines and from on-premises to Google Cloud.
- 4.Domain 4 – Deploy scalable and highly available databases in Google Cloud (~20%): Covers provisioning database instances and clusters; configuring read replicas and multi-region replication; setting up and periodically testing HA failover and disaster recovery procedures; configuring network connectivity and Private Service Connect; implementing encryption at rest and in transit; and validating deployment configurations against performance and availability requirements.
Study Tips
- Start with the official Professional Cloud Database Engineer exam guide (PDF available at services.google.com/fh/files/misc/professional_cloud_database_engineer_exam_guide_english.pdf) and map every subtopic to your existing knowledge gaps before building a study schedule.
- Complete Google's official Cloud Database Engineer Learning Path on Cloud Skills Boost (cloudskillsboost.google/paths/22), which includes structured courses, hands-on Qwiklabs, and skill badges covering Cloud SQL, Spanner, Bigtable, and Firestore.
- Practice database selection trade-offs extensively — the exam frequently presents scenarios requiring you to choose between Cloud SQL, AlloyDB, Cloud Spanner, and Bigtable based on consistency, scalability, latency, and cost constraints. Build a comparison matrix for each product.
- Use Google's official sample questions form (linked from the certification page) to calibrate your readiness, then focus remediation on Domain 1 (database design, ~32%) and Domain 3 (migration, ~23%), which together represent over half the exam weight.
- Get hands-on experience with Database Migration Service and Datastream by completing migration labs — migration scenarios are heavily tested and require knowing tool selection, minimal-downtime strategies, and schema conversion specifics that are difficult to learn from reading alone.
- Study IAM roles, VPC configuration, Private Service Connect, and encryption options for each database product, as security and access control questions appear across multiple domains.
- Review the Cloud Monitoring and Query Insights documentation for Cloud SQL and Spanner, and practice interpreting performance metrics and identifying bottlenecks — operational management and troubleshooting questions are prominent in Domain 2.
Career Benefits
Earning the Professional Cloud Database Engineer certification signals to employers a validated, senior-level ability to architect and operate Google Cloud database infrastructure — a skill set in high demand as enterprises migrate legacy database workloads to managed cloud services. Certified professionals are well-positioned for roles such as Cloud Database Engineer, Cloud Infrastructure Architect, Data Platform Engineer, and Site Reliability Engineer with a database specialization.
Google Cloud certifications at the Professional level are widely recognized in the industry and often listed as preferred or required qualifications in job postings at companies running workloads on Google Cloud. While Google does not publish salary data tied to this specific certification, cloud database engineering roles commanding this skill set typically command competitive salaries in line with other Google Cloud Professional-level certifications. The credential complements adjacent certifications such as the Professional Data Engineer and Professional Cloud Architect, and can serve as a differentiator for professionals seeking to specialize in the growing field of cloud-native database management.
Sample Questions
Preview — answers shown5 sample questions with correct answers and explanations. Start a practice session to test yourself across all 608 questions.
1. An e-commerce application uses Memorystore for Redis Standard tier. During flash sales, cache hit rates drop from 95% to 60% due to new product pages, causing database overload. Memory utilization is at 50%. What configuration change improves cache hit rates during flash sales?
AImplement cache warming by pre-loading product data before flash sales
BChange eviction policy from allkeys-lru to volatile-lru
CUpgrade to a larger Memorystore instance with more memory
DEnable read replicas to distribute read load
Explanation
Cache warming pre-loads product data into Redis before the flash sale begins, ensuring high cache hit rates when traffic arrives. Since memory utilization is only 50%, capacity is not the issue. Upgrading instance size doesn't solve the cold cache problem. Changing eviction policy to volatile-lru only evicts keys with TTL set, which may not be appropriate for all cached data. Read replicas help with throughput but don't improve hit rates - the data still needs to be cached. Cache warming is a standard practice for predictable traffic spikes, proactively loading data to avoid cache misses during critical periods.
2. Your application uses Firestore in Datastore mode for a mobile game. Players can have thousands of achievements. You need to efficiently query for players who have earned a specific achievement. How should you model the data?
AStore achievements as an array property in the Player entity and create a composite index on the array
BStore all achievements in a single JSON property in the Player entity
CCreate a PlayerAchievements entity with an array of achievement IDs and use array-contains queries
DCreate a separate Achievement entity for each player-achievement combination with player_id and achievement_id properties
Explanation
Creating separate Achievement entities with player_id and achievement_id enables efficient querying using composite indexes and avoids array size limitations. This pattern scales well with thousands of achievements per player and supports fast lookups. Arrays in Firestore (A, D) have size limitations and can cause entity size issues. JSON properties don't support efficient querying and require client-side filtering.
3. A data warehouse migration involves loading 100 TB of historical sales data into BigQuery. The data is currently in 50,000 CSV files in Cloud Storage. What is the most efficient and cost-effective loading strategy?
AImplement Cloud Dataflow pipelines to transform and stream data into BigQuery
BCreate a Compute Engine VM to sequentially load files using bq load commands
CUse BigQuery Data Transfer Service to schedule incremental loads over several weeks
DUse BigQuery Load Jobs with wildcard URIs to load all files in parallel automatically
Explanation
BigQuery Load Jobs natively support wildcard patterns (gs://bucket/*.csv) and automatically parallelize loading across multiple files. For batch loads from Cloud Storage, this is the most efficient method with no compute costs (load jobs are free except for storage and queries). BigQuery optimally distributes the workload. Dataflow is unnecessary complexity and incurs compute costs for simple loads without transformation. Sequential loading from a VM is slow and inefficient. Data Transfer Service is designed for scheduled ongoing transfers, not one-time bulk loads.
4. A media streaming application uses Firestore to store user preferences and viewing history. The application experiences 10,000 document writes per second during peak hours. You notice write latency increasing and some write operations timing out. What is the most likely cause and solution?
AFirestore instance size limit; migrate to Bigtable for higher throughput
BHot spotting on document IDs; implement distributed counters or sharding
CInsufficient indexes; create composite indexes for write operations
DNetwork bandwidth limitation; upgrade to premium networking tier
Explanation
Firestore has a limit of 1 write per second per document. If writes are concentrated on a small number of documents (hot spotting), those documents become bottlenecks. Distributed counters or sharding distributes writes across multiple documents. Firestore doesn't have instance size limits - it scales horizontally. Network bandwidth is rarely the bottleneck for document writes. Indexes affect query performance but can actually slow writes slightly, not solve write bottlenecks. Understanding Firestore's per-document write limits is critical for designing high-throughput applications. Sharding patterns distribute load to avoid single-document bottlenecks.
5. You need to provision infrastructure for a new microservices application that will use 15 different Cloud SQL databases. Each service connects to its dedicated database. What provisioning approach optimizes for cost and operational simplicity? (Select two!)
Multiple correct answersACreate a single Cloud SQL instance with 15 databases for all services
BDeploy 15 individual Cloud SQL instances, one per service
CUse Terraform modules to standardize and automate instance provisioning
DCreate shared-core instances (db-f1-micro) for low-traffic services
EImplement a service mesh to reduce database connection overhead
FUse Cloud SQL Proxy to multiplex connections across instances
Explanation
A single Cloud SQL instance with multiple databases reduces infrastructure costs while providing logical isolation between services. Terraform modules standardize configuration and automate deployment, reducing operational complexity for managing 15 databases. Individual instances per service multiply costs unnecessarily when workloads can be consolidated. Shared-core instances are cost-effective for very light workloads but the question does not indicate this is appropriate for all services. Service mesh addresses service-to-service communication but does not reduce database costs. Cloud SQL Proxy handles connectivity and security but does not reduce infrastructure costs. The combination of consolidated infrastructure and infrastructure-as-code automation provides the best balance of cost optimization and operational simplicity.
$7
One-time access to this exam
Full access to all 608 questions
Or $15/mo for all 201 exams
Detailed explanations
Free preview stays available