Prepare and test your skills

Prepare and test your skills

Object storage is a cloud storage architecture that manages data as distinct, immutable units called objects. In Google Cloud, Cloud Storage is the scalable, managed object storage service that keeps objects in containers called buckets. Each object is an immutable file of any format, bundled with metadata and a unique identifier used to locate and access it, and each bucket can hold essentially unlimited objects. Buckets also define storage class and location, which control durability and availability choices. Object storage of this type is designed for large amounts of unstructured data, including websites, streaming video, and mobile apps.
Cloud Storage organizes resources in a fixed hierarchy that flows from organization to project to bucket to object. All buckets belong to a project, projects are grouped under an organization, and each project has its own set of Cloud Storage APIs. Resource names make this hierarchy explicit as projects/_/buckets/BUCKET_NAME for buckets and projects/_/buckets/BUCKET_NAME/objects/OBJECT_NAME for objects, where a trailing #NUMBER indicates a generation and #0 is the most recent version.
Within a bucket, folders and managed folders add directory-like control in different ways. Folders exist only in buckets with hierarchical namespace enabled and provide a real file system, so a folder and all objects inside it can be renamed atomically in one operation. Managed folders do not use a true directory tree; a managed folder is a resource overlay used only for permission checking that grants or revokes additional access beyond the Identity and Access Management (IAM) permissions set on the bucket.
Buckets also act as a primary data foundation for other Google Cloud services. Buckets can be connected as storage backends for services such as AI Hypercomputer, Vertex AI, and Google Kubernetes Engine (GKE), so analytics and AI workloads can read directly from object storage. You can work with buckets and objects through the Google Cloud console, the Google Cloud CLI, or client libraries in C++, C#, Go, Java, Node.js, PHP, Python, and Ruby.
Object storage in Cloud Storage achieves durability and global accessibility through bucket location types and replication. A regional bucket replicates data synchronously across zones inside one region. Dual-region and multi-region buckets store data redundantly in at least two separate geographic locations, with metadata written synchronously across regions and data replicated asynchronously. For dual-region buckets, turbo replication provides replication across region pairs with a recovery point objective (RPO) of 15 minutes. Access can be limited to principals you specify or opened to everyone on the public internet, which combines replication with flexible sharing for global use.
A top-down hierarchy from organization to project to bucket to object, showing folders as a real file system and managed folders as a permission overlay.
| Bucket location type | Replication scope | Replication behavior |
|---|---|---|
| Regional | Zones within one region | Data replicated synchronously across zones in the region |
| Dual-region | At least two separate geographic locations | Metadata written synchronously across regions; data replicated asynchronously; supports turbo replication with 15-minute RPO |
| Multi-region | At least two separate geographic locations | Metadata written synchronously across regions; data replicated asynchronously |
Object storage is the best fit when data is large-scale, unstructured, immutable, and needs to be durable and globally accessible. It supports analytics and AI/ML workloads because buckets serve as shared storage backends for processing engines, rather than acting as block storage for virtual machine (VM) disks. It also supports disaster recovery (opens in a new tab), backup, and business continuity because replicated buckets keep copies available across zones or regions. Common uses include:
For long-term retention and cost control, storage classes, locations, and lifecycle management let you match durability and availability to the workload. Choose object storage for analytics, archival, and large-scale unstructured data when you need managed, scalable storage for objects rather than disks attached to VMs.
Exam tip: Object storage keeps immutable objects in buckets, while block storage provides network-based disks that VMs and containers use like physical disks.
A relational database is a system that stores highly structured data in tables of rows and columns. Each table holds one kind of record, each row is one record, and each column is one data point, such as a patient table where each row is one patient and columns hold name, insurance information, and contact details. A relational database management system (RDBMS) (opens in a new tab) maintains links across multiple tables, so related data elements can be connected without storing the same data multiple times. This row-based table structure makes it easy to gain insights about relationships between columns and data points. Tables can also use indexes on particular columns for faster lookups. Relational systems in this category include PostgreSQL, SQL Server, MySQL, and Oracle Database.
Structured Query Language (SQL) is the programming language used to store, retrieve, manage, and query data in a relational database. SQL uses English-like statements such as CREATE to create a table, INSERT to add a new row, and SELECT to query data. Statements that make structural or administrative changes are usually reserved for developers and administrators, while end-user applications perform read and write operations. Although created for relational databases, SQL now underlies many technologies and has different variants by database system. In BigQuery, the ANSI-compliant dialect GoogleSQL groups work into distinct statement types:
Relational storage works best when data is highly structured, relationships between tables matter, and workloads need consistent, connected records such as transaction processing. No single database fits every project, so the choice depends on the use case and the need to link related data without duplication. On Google Cloud, Cloud SQL is the fully managed relational service for MySQL, PostgreSQL, and SQL Server that handles setup, maintenance, patches, backups, and replication.
| Engine | Ownership and type | Key facts |
|---|---|---|
| MySQL | Open-source relational database created in 1995 and sponsored by Oracle | Supports SQL queries through graphical interface or command line; can be deployed manually or through a managed cloud service |
| PostgreSQL | Open-source, object-relational database | Valued for enterprise performance, reliability, and robust features; supports procedural extensions including Python, Perl, Tcl, Java, Ruby, and C++ |
| SQL Server | Microsoft-owned RDBMS built on SQL | Uses row-based tables to connect related data; supports business intelligence, transaction processing, data analytics, and machine learning |
Exam tip: PostgreSQL is open source, while SQL Server is owned by Microsoft — both use SQL, but ownership is the major difference to remember.
Non-relational databases (opens in a new tab), also called NoSQL databases, store data without forcing every record into fixed tables, rows, and columns. When data outgrows the limits of a traditional relational model, these Google Cloud services are designed to take over. They increase flexibility because records do not have to match one rigid schema, and they scale automatically as data volume or user count grows. Google Cloud labels its non-relational services by data model, and each label signals the shape of data it handles best:
Firestore (opens in a new tab) is the fully managed, scalable, and serverless document database for rich applications on mobile, web, and server. Firestore is flexible and scalable for application development, and Firestore with MongoDB compatibility lets applications already written against MongoDB use a Firestore-based service. Choose a document model when the problem is flexible application data with changing fields, rather than massive key-driven volume. Datastore is a second NoSQL document database built for automatic scaling, high performance, and ease of application development. Google Cloud lists Datastore as a highly scalable NoSQL database for web and mobile applications, and its comparison table places it under the NoSQL: Indexed category. Firestore and Datastore therefore share the document model and automatic-scaling goal, while other non-relational services serve different shapes of data.
Bigtable is the fully managed, scalable NoSQL service for large analytical and operational workloads that store terabytes or petabytes of data. Bigtable uses a NoSQL wide-column format and is also listed under the NoSQL: Key-value category, so it fits datasets where volume is very large and access is driven by keys rather than document structure. Data flows from applications to Bigtable by key lookup, which supports high-throughput operational reads and writes as well as large analytical scans. Choose Bigtable when volume is the defining problem. Choose Firestore or Datastore when the defining problem is flexible application data for web and mobile development.
A decision tree that branches on latency, data volume, and app needs to select Memorystore, Bigtable, Firestore, or Datastore.
Memorystore is the scalable, secure, and highly available in-memory service that reduces latency for applications running on Google Cloud. Data is held in memory rather than on disk, so applications that need extreme performance can read results much faster. Memorystore is offered as Memorystore for Valkey, Memorystore for Redis Cluster, and Memorystore for Redis, each providing a managed service for Valkey or Redis workloads. Memorystore for Memcached is also listed as an in-memory option but is marked deprecated, so new designs should favor the Valkey or Redis offerings. All of these non-relational services are fully managed, which means Google Cloud handles operational work rather than the application team.
The non-relational services differ by data model, target scale, and best fit. The table below preserves the model labels and scale differences that determine which service to choose.
| Service | Data model | Best fit |
|---|---|---|
| Bigtable | NoSQL wide-column, also listed as key-value | Terabytes to petabytes in large analytical and operational workloads |
| Firestore | Document | Rich mobile, web, and server applications, including MongoDB-compatible workloads |
| Datastore | NoSQL document, listed as indexed | Web and mobile applications needing automatic scaling and high performance |
| Memorystore | In-memory | Applications needing reduced latency through Valkey or Redis |
Exam tip: Bigtable is the wide-column, petabyte-scale NoSQL service, while Firestore and Datastore are document databases for application development.
Object storage is a cloud storage architecture that manages data as distinct, immutable units called objects, and in Google Cloud, Cloud Storage keeps objects in containers called buckets. Each object is an immutable file of any format bundled with metadata and a unique identifier, and resources follow a fixed hierarchy from organization to project to bucket to object. Buckets also define storage class and location, which control durability and availability choices.
Object storage is the best fit when data is large-scale, unstructured, immutable, and needs to be durable and globally accessible, including analytics, archival, backup, disaster recovery, and content distribution. Buckets serve as shared storage backends for processing engines rather than acting as block storage for virtual machine disks. Block storage provides network-based disks that virtual machines and containers use like physical disks.
A relational database stores highly structured data in tables of rows and columns and uses Structured Query Language to store, retrieve, manage, and query data, with links across tables so related data can be connected without duplication. Non-relational databases, also called NoSQL databases, store data without forcing every record into fixed tables, rows, and columns, which increases flexibility and scales automatically as data volume or user count grows. On Google Cloud, Cloud SQL is the fully managed relational service for MySQL, PostgreSQL, and SQL Server, while Bigtable, Firestore, Datastore, and Memorystore are non-relational services.
Choose Bigtable when volume is the defining problem because it uses a NoSQL wide-column format for large analytical and operational workloads that store terabytes or petabytes of data accessed by key lookup. Choose Firestore or Datastore when the defining problem is flexible application data for web and mobile development, since both are document databases built for automatic scaling. Choose Memorystore when the defining need is reduced latency for extreme performance, since data is held in memory through Valkey or Redis offerings.
A healthcare organization is designing a new application to manage patient records, insurance policies, and billing transactions. The system must meet the following requirements:
SELECT and INSERTWhich data management solution should the organization implement?
Last updated