Cloud Storage is Google Cloud's object storage service. It holds all data types, which is why it is often described as blob storage, and it is the go-to home for unstructured data such as text files, images, videos, and large datasets. For someone working with databases, the most common reason to reach for it is backups, so it shows up frequently on the Professional Cloud Database Engineer exam in scenarios about where to put database exports, archives, and snapshots. The service organizes data into containers called buckets, and each individual file or item inside a bucket is called an object. If you have worked with AWS S3, the model is the same, including the term bucket.
A bucket is the top-level container, and an object is a single file stored inside it. Cloud Storage can accommodate any format and any data transfer type, whether you are moving a bulk database export, raw files, or analytics datasets. Access can be controlled at the bucket level and also down to specific objects within a bucket, which lets you share some data with certain users while keeping the rest restricted. Two features that matter for data protection are versioning and redundancy. Versioning keeps multiple versions of the same object, so you can recover from an accidental deletion or an unwanted change, and redundancy replicates data across locations so it survives a failure. Cloud Storage is generally cheaper than most other storage options, which is part of why it is a practical place for backups, archives, and logs that are not accessed often but still need to be available.
The service is well suited to a range of backup needs. Archives are long-term datasets that do not require frequent access. Database backups come in many formats and sizes, and Cloud Storage can hold any of them. Media backups such as photos and videos benefit from large-scale capacity and global accessibility. Log files are often generated in high volume, and they benefit from centralized management and lifecycle policies that automate retention and deletion. Disk snapshots, usually taken from Compute Engine instances, are also commonly kept in Cloud Storage. The common thread for the exam is that when a question asks where durable, cost-conscious backup or archive data should live, Cloud Storage is usually the intended answer.
Storage classes let you match the cost of storage to how often you actually read the data. Standard is for data you access frequently. Nearline and Coldline are progressively cheaper to store but are meant for data you access less often, with Coldline aimed at colder data than Nearline. Archive is the lowest-cost class and is intended for data you rarely touch but still need to keep available. The general pattern to keep in mind is a trade-off: the colder the class, the lower the storage cost, but the more it is designed around infrequent access rather than constant reads. For database backups and archives that you keep for compliance or recovery but seldom open, the colder classes are typically the economical choice, while data you query or restore regularly belongs in Standard.
Rather than changing storage classes by hand as data ages, you can define lifecycle policies on a bucket and let Cloud Storage manage the transitions automatically. A lifecycle policy is a set of rules that can move objects to a colder class after a period of time or delete them once they are no longer needed. This is what makes the service practical for high-volume backups and logs, because retention and deletion become automatic rather than a manual chore. On the exam, when a scenario describes data that should move to cheaper storage as it ages or be removed after a retention window, lifecycle management is the mechanism being tested.
Where a bucket lives is a separate decision from its storage class, and Cloud Storage gives you three location types. Regional storage keeps data in a single geographic region. It is the right choice when you need data to stay in one location for compliance, when you want low-latency access from resources in that same region, and when cost matters, since regional is generally the cheapest of the three. Multi-region storage spans a large area such as the US, the EU, or Asia. It offers the highest availability and global access with minimal latency, so it fits workloads that need broad, often continent-level, reach. Dual-region storage sits in between. It stores data across two distinct regions, which gives higher durability and availability than a single region while costing less than multi-region, and that pairing of two separate locations makes it a good fit for disaster recovery.
For a database engineer, the location choice usually comes down to balancing three things: keeping data in a required jurisdiction, getting it close to the systems that read it, and surviving the loss of a location. Regional answers the first two cheaply, dual-region adds resilience across two places for disaster recovery, and multi-region maximizes availability and reach. Many exam questions are really asking you to separate these properties and pick the option whose trade-offs match the scenario.
The command line tool for the service is gcloud storage. It handles the common operations such as uploading, downloading, deleting, and copying files, and it can enable parallel composite uploads to make better use of available bandwidth on large files. The commands you reach for most are gcloud storage cp to copy, gcloud storage rsync to sync directories or buckets, and gcloud storage ls to list bucket contents. The older tool gsutil still works, but Google now recommends gcloud storage instead, so that is the one to be familiar with.
gcloud storage cp backup.sql gs://my-db-backups/
gcloud storage rsync ./exports gs://my-db-backups/exports
gcloud storage ls gs://my-db-backups/Our Professional Cloud Database Engineer course covers Cloud Storage alongside storage classes and lifecycle policies, with practice questions that drill these distinctions.
