Planning and implementing a cloud solution · Topic 5 of 9
Choosing Data and Storage Products
Choosing Cloud SQL, AlloyDB, Spanner, Bigtable, Firestore, BigQuery, Memorystore, Pub/Sub, Kafka, Filestore and Cloud Storage classes for ACE.
- 4 min
- Cloud SQL (standard relational), AlloyDB (high-performance PostgreSQL), Spanner (global relational), Bigtable (wide-column, huge scale), Firestore (documents, mobile), BigQuery (analytics), Memorystore (cache).
- Pub/Sub (messaging), Managed Kafka (Kafka compatibility), Dataflow (pipelines).
- Filestore (NFS), NetApp Volumes (NFS and SMB, enterprise), Managed Lustre (HPC and AI throughput), Cloud Storage (objects).
What it is
Google Cloud has a data product for each kind of workload: relational and NoSQL databases, a data warehouse, caches, messaging and stream processing, file storage and object storage. The exam gives requirements (data model, scale, latency, consistency, protocol) and asks which product to choose and deploy.
How it works
Databases and analytics
| Product | Type | Choose when |
|---|---|---|
| Cloud SQL | Managed MySQL, PostgreSQL, SQL Server | Standard relational apps, lift and shift of these engines, regional scale |
| AlloyDB for PostgreSQL | High-performance PostgreSQL-compatible | Demanding PostgreSQL workloads, mixed transactional and analytical queries |
| Spanner | Relational, horizontally scalable, strongly consistent, regional or multi-region | Global scale, very high availability, relational with no sharding by hand |
| Bigtable | Wide-column NoSQL | Huge throughput with low latency: time series, IoT, ad tech, key-based lookups |
| Firestore | Serverless document database | Mobile and web apps, flexible schemas, real-time sync, offline support |
| BigQuery | Serverless data warehouse | SQL analytics over large datasets, BI, BigQuery ML |
| Memorystore | Managed Redis, Valkey or Memcached | In-memory caching and session stores for very low latency |
Messaging and processing
| Product | What it does | Choose when |
|---|---|---|
| Pub/Sub | Global, serverless messaging with at-least-once delivery | Decouple services, ingest event streams |
| Google Cloud Managed Service for Apache Kafka | Managed Apache Kafka clusters | Existing Kafka applications and tooling, Kafka protocol compatibility |
| Dataflow | Serverless Apache Beam for batch and streaming pipelines | Transform and enrich data in motion, load into BigQuery |
File and object storage
| Product | What it is | Choose when |
|---|---|---|
| Cloud Storage | Object storage in buckets | Unstructured data, backups, data lakes, static website content |
| Filestore | Managed NFS file shares | Applications needing a shared POSIX file system, such as content management or GKE workloads |
| Google Cloud NetApp Volumes | Managed NetApp file storage (NFS and SMB) | Enterprise file workloads, including Windows SMB shares, and NetApp features |
| Google Cloud Managed Lustre | Managed parallel file system | High-performance computing and AI training needing very high throughput |
Cloud Storage classes
| Class | Access pattern | Minimum storage duration |
|---|---|---|
| Standard | Frequently accessed | None |
| Nearline | About once a month | 30 days |
| Coldline | About once a quarter | 90 days |
| Archive | Less than once a year | 365 days |
Colder classes cost less to store and more to retrieve. Autoclass moves objects between classes automatically based on access. Location choice (region, dual-region, multi-region) affects availability and latency.
gcloud storage buckets create gs://shop-backups-1234 \
--location=us-central1 --default-storage-class=NEARLINE --uniform-bucket-level-accessLook for the deciding words: "global" and "strongly consistent relational" mean Spanner; "PostgreSQL, high performance" means AlloyDB; "millisecond lookups at massive scale" means Bigtable; "shared NFS file system" means Filestore.
Deploying a database
gcloud sql instances create orders-db --database-version=POSTGRES_16 \
--edition=ENTERPRISE --region=us-central1 --tier=db-custom-2-7680 --availability-type=REGIONAL--availability-type=REGIONAL creates a high-availability Cloud SQL instance with a standby in another zone.
A worked scenario
A gaming company needs player profiles in a mobile app with offline sync (Firestore), a global leaderboard ledger that must be consistent worldwide (Spanner), session caching (Memorystore), telemetry of millions of events per second with fast lookups (Bigtable), event ingestion (Pub/Sub), analytics (BigQuery), a shared file system for build servers (Filestore), and game assets in Cloud Storage Standard, with old builds in Archive.
How the exam asks about this
- Requirements are listed and you choose the product.
- "Existing Kafka producers and consumers, minimal code change." Managed Service for Apache Kafka.
- "Shared NFS storage for several VMs." Filestore.
- "Data read once a quarter for compliance." Coldline.
What to remember for the exam
- Cloud SQL (standard relational), AlloyDB (high-performance PostgreSQL), Spanner (global relational), Bigtable (wide-column, huge scale), Firestore (documents, mobile), BigQuery (analytics), Memorystore (cache).
- Pub/Sub (messaging), Managed Kafka (Kafka compatibility), Dataflow (pipelines).
- Filestore (NFS), NetApp Volumes (NFS and SMB, enterprise), Managed Lustre (HPC and AI throughput), Cloud Storage (objects).
- Storage classes: Standard, Nearline 30, Coldline 90, Archive 365 days minimum.
BigQuery is for analytics, not for an application's frequent small transactional updates.
Common mistakes
Choosing Bigtable for relational queries with joins. It is NoSQL.
Choosing Cloud SQL for global, multi-region writes. Spanner fits.
Using Cloud Storage as a shared POSIX file system. Use Filestore or NetApp Volumes.
Check yourself
Which database provides globally consistent, horizontally scalable relational storage?
Spanner.
Which product provides managed NFS file shares?
Filestore.
What is the minimum storage duration for Coldline?
90 days.
Topic 5 of 9 in Planning and implementing a cloud solution
Test yourself on Planning and implementing a cloud solution
The full Google Cloud Associate Cloud Engineer question bank has 661 questions, with timed practice exams and an explanation for every answer.
This study guide is independent and is not affiliated with, endorsed by, or sponsored by Google Cloud. Vendor names, certification names and exam codes are trademarks of their respective owners and are used for identification only.