Associate Data Practitioner

Choose the Appropriate Data Storage Location Type

Selecting the right storage location type determines how resilient, fast, and cost-effective your data storage will be. Google Cloud offers several location options that define the physical boundaries where your data is stored and replicated. Choosing the correct option depends on where your users are located, how much latency they can tolerate, and your disaster recovery requirements.

Zonal storage places your data within a single, specific zone inside a geographic region, which is common for persistent disks attached to virtual machines. This option offers very low latency for local compute resources, but it lacks protection if that entire zone experiences an outage. To protect against localized disasters, regional storage replicates your data across multiple zones in a single region, which is the default choice for services like Cloud SQL to ensure high availability.

For even greater resilience, dual-regional and multi-regional storage options replicate your data across wider geographic areas. Dual-regional storage mirrors your data across two specific regions, which helps balance fast access for users in those areas with protection against a region-wide outage. Multi-regional storage distributes data across at least three geographic regions, making it the ideal choice for Cloud Storage buckets that serve a global audience or require maximum durability.

Classify Use Cases by Data Requirements

Before selecting a storage service, you must classify your data into structured, semi-structured, or unstructured requirements. Each data type has a unique organization style, and matching your data to the correct storage system ensures efficient querying and lower costs. Google Cloud provides specialized database and storage services designed specifically for each classification.

Structured data has a highly organized, predictable format that fits perfectly into tables with rows and columns. Choose Cloud SQL when you need a traditional relational database to handle transactional operations within a single region with strict consistency. If your transactional application must scale globally across multiple regions, choose Spanner to maintain that strict consistency without sacrificing performance. For heavy analytics and business intelligence, choose BigQuery because it acts as an analytical data warehouse designed to query massive structured datasets.

Semi-structured data does not fit into a rigid table but still contains markers, tags, or hierarchies to separate the information. Choose Firestore when you need a flexible, document-based NoSQL database that can store user profiles and automatically sync data across mobile devices. When your application processes rapid, continuous streams of data like time-series or Internet of Things (IoT) events, choose Bigtable because it can scale to millions of reads and writes with sub-millisecond latency.

Unstructured data lacks any pre-defined organization, which means it consists of files like images, videos, audio clips, and system backups. For this type of data, choose Cloud Storage because it is a highly scalable object storage service that can hold virtually unlimited files of any size. Within Cloud Storage, you organize files into containers called buckets, where you can easily apply security policies and manage the lifecycles of your unstructured files.