- A cloud database runs on a provider's infrastructure; in the managed form the provider also handles installation, patching, backups and failover, while you keep schema design, queries and cost control.
- For a typical application database, start with a managed relational service (Amazon RDS, Azure SQL Database or Azure Database for PostgreSQL, Google Cloud SQL) or a specialist PostgreSQL or MySQL platform.
- Serverless databases suit idle or spiky workloads because compute can pause; NoSQL services suit key-value and document access patterns; warehouses are for analytics, not application traffic.
- Choose by engine first, then by the cloud your application already runs on, then by pricing model and the operational features you need (high availability, backups, regions).
What is a cloud database?
A cloud database is a database that runs on a cloud provider's infrastructure and that you reach over the network, instead of a database installed on a server you own. In practice the term almost always means a managed service: you choose an engine and a size, and the provider runs the servers, the operating system and the database software for you.
The difference from simply renting a virtual machine is who does the operational work. AWS's own documentation for Amazon RDS describes it as "a managed database service" that is "responsible for most management tasks", and compares it with running a database on an EC2 virtual machine, where scaling, high availability, backups, patching and installation stay with the customer. Microsoft describes Azure SQL Database as a fully managed platform as a service that handles upgrading, patching, backups and monitoring without user involvement. Google describes Cloud SQL as a fully managed relational database service for MySQL, PostgreSQL and SQL Server.
What you still own is the part that depends on your application: the schema, the queries and indexes, access control, and the bill. AWS states this directly: you are responsible for query tuning. Good indexing and sensible transaction design matter as much on a cloud database as on your own server.
Types of cloud database
Most cloud database services fall into five groups. They overlap (a serverless database is also managed, and many DBaaS platforms run on the big clouds), but the groups help you narrow the choice.
| Type | What it is | Examples | Typical fit |
|---|---|---|---|
| Managed relational | A standard SQL engine run for you on instances you size and pay for by the hour | Amazon RDS, Azure SQL Database, Azure Database for PostgreSQL, Google Cloud SQL | Most web and business applications |
| Serverless relational | Compute scales with load and can pause when idle; you pay for usage | Aurora Serverless, Azure SQL Database serverless, Neon | Development databases, intermittent or spiky traffic |
| DBaaS platforms | Specialist providers offering one or a few engines with their own tooling | Supabase, Neon, PlanetScale, Aiven, DigitalOcean, MongoDB Atlas | Teams that want a simpler console, branching or multi-cloud |
| NoSQL | Key-value, document or multi-model databases with their own APIs | Amazon DynamoDB, Azure Cosmos DB, Firestore, MongoDB Atlas | High-volume key lookups, flexible documents, mobile backends |
| Data warehouse | Columnar engines for analytical queries over large data | Amazon Redshift, Google BigQuery, Snowflake | Reporting and analytics, not application transactions |
Managed relational databases
This is the default choice for an application that needs SQL, joins and transactions. You pick an engine (PostgreSQL, MySQL, MariaDB, SQL Server, Oracle or Db2 on Amazon RDS), an instance size and storage, and the provider handles the rest. Our managed databases guide covers what the provider does and what stays with you.
Serverless databases
Serverless services separate the database from fixed instance sizes. AWS describes Aurora Serverless as an on-demand, autoscaling configuration that starts up, shuts down and scales capacity based on the application's needs. Azure SQL Database serverless bills compute per second and pauses during inactive periods, when only storage is billed. Neon suspends a Postgres compute after 5 minutes of inactivity. The trade-off is that a paused database takes time to resume. See serverless databases explained.
DBaaS platforms and specialist providers
Database as a service (DBaaS) is the general model behind all of these offerings, but the term is often used for providers that focus on databases alone. Supabase gives every project its own dedicated Postgres instance with a backend platform around it; Neon separates Postgres compute from storage; MongoDB Atlas runs MongoDB on AWS, Google Cloud and Azure; Aiven runs open source engines across several clouds. Our DBaaS guide explains the service model and how to evaluate a provider.
NoSQL and data warehouses, briefly
NoSQL cloud databases trade SQL features for a specific access pattern and scale model. AWS describes DynamoDB as a serverless, fully managed, distributed NoSQL database; Microsoft describes Azure Cosmos DB as a fully managed NoSQL and vector database. If you are unsure which model fits, read SQL vs NoSQL and DynamoDB vs PostgreSQL.
Warehouses such as Amazon Redshift and BigQuery (which Google calls a serverless data analytics platform) are built for analytical queries over large tables. They sit next to an application database rather than replacing it; see data warehouse vs database.
Cloud database providers: the big three and the specialists
Three hyperscale clouds offer the widest range of database services, and a group of specialist providers competes on developer experience, pricing model or a single engine. The table lists the main relational and NoSQL services by provider; it is not exhaustive.
| Provider | Managed relational | Serverless or scale-to-zero | NoSQL and other |
|---|---|---|---|
| AWS | Amazon RDS (PostgreSQL, MySQL, MariaDB, SQL Server, Oracle, Db2), Amazon Aurora | Aurora Serverless (Aurora PostgreSQL and Aurora MySQL) | DynamoDB; Redshift for analytics |
| Microsoft Azure | Azure SQL Database, Azure Database for PostgreSQL, Azure Database for MySQL | Azure SQL Database serverless compute tier | Azure Cosmos DB |
| Google Cloud | Cloud SQL (MySQL, PostgreSQL, SQL Server), AlloyDB for PostgreSQL, Spanner | Not a Cloud SQL feature; Firestore and BigQuery are serverless | Firestore; BigQuery for analytics |
| Specialists | Supabase, DigitalOcean, Aiven, PlanetScale (MySQL-compatible Vitess and Postgres) | Neon (Postgres), Cloudflare D1 (SQLite semantics) | MongoDB Atlas |
When the big three make sense
If your application already runs on AWS, Azure or Google Cloud, the same provider's database keeps traffic inside one network, one bill and one identity system. The hyperscalers also offer the most regions and the widest engine choice, including commercial engines (SQL Server, Oracle) that most specialists do not run. For side-by-side detail, see AWS RDS vs Azure SQL Database, AWS RDS vs Google Cloud SQL and Azure SQL vs Google Cloud SQL.
When a specialist makes sense
Specialists usually win on simplicity and on features the big clouds handle differently: database branching per pull request, scale to zero on small databases, all-inclusive monthly plans, or a backend platform around the database. The trade-offs are fewer regions, fewer engines and, in some cases, a newer company or a recent change of ownership. Comparisons such as Supabase vs Neon, Neon vs AWS RDS and PlanetScale vs AWS RDS go into those differences.
How to choose a cloud database
Work through these questions in order. Each one removes options, and by the end there are usually two or three services worth pricing.
- Data model and engine. Relational data with joins and transactions points to PostgreSQL, MySQL or SQL Server. Pick the engine your application and team already know; changing engine is a bigger project than changing provider.
- Where the application runs. A database in the same cloud and region as the application avoids cross-provider latency and data transfer charges.
- Workload shape. Steady traffic suits an always-on instance billed by the hour. Long idle periods suit a serverless option that pauses compute. Analytics over large tables suits a warehouse.
- Operational needs. Decide what you need for high availability (a standby in another zone), backup retention, point-in-time restore, regions and compliance, then check which plans include them.
- Pricing model. Compare like for like: instance hours, storage, backups and data transfer. Our cloud database pricing comparison works through the same small workloads on AWS, Azure and Google Cloud.
- Exit route. Standard engines can be dumped and restored elsewhere; proprietary APIs and platform features (auth, functions, NoSQL query languages) need rewriting if you leave.
For the infrastructure decision behind all of this (shared hosting, a VPS, a managed service or serverless), read database hosting: how to choose. For a shortlist of providers with criteria and limitations, see best database hosting providers.
Free and low-cost cloud databases for small projects
Several providers offer a free way to start, but the terms differ a lot, and the difference between an always-free tier and a time-limited trial matters if you plan to keep the database. Azure SQL Database has a free offer that Microsoft describes as available per database for the lifetime of the subscription, with monthly limits on vCore seconds and storage. Neon and Supabase have free plans with no time limit but with storage caps and rules for idle projects (Supabase pauses free projects after a week of inactivity). Google's Cloud SQL free trial instance lasts up to 30 days. AWS offers new customers a Free plan for up to 6 months on selected RDS instance types.
For a small business application, the cheapest option that will stay running is rarely a free tier: plan for a small paid instance with automated backups, and use free tiers for learning and prototypes. The free database hosting comparison lists the documented limits side by side.
Common mistakes when moving a database to the cloud
- Comparing instance prices only. Storage, backups beyond the free allowance, standby replicas and outbound data transfer are billed separately on most services.
- Assuming managed means tuned. The provider keeps the server running; slow queries and missing indexes remain your job.
- Ignoring idle behaviour. A free or serverless database that pauses can add a delay to the first request after a quiet period. Check what the provider documents before using one for latency-sensitive traffic.
- Picking a region by habit. Put the database in the same region as the application, and check that the region offers the engine version and high-availability option you need.
- No exit plan. Keep regular logical exports (for example
pg_dumpormysqldump) so moving provider stays possible.
All cloud database guides
Cloud database basics
- Managed Databases Explained: How They Work, Costs & Providers
- Database as a Service (DBaaS): What It Is & Top Providers
- Database Hosting: How to Choose the Right Option (Complete Guide)
- Best Database Hosting Providers in 2026
- Best Free Database Hosting in 2026 (Limits Compared)
- Serverless Databases Explained: How They Work & When to Use Them
- Cloud Database Pricing Compared: AWS vs Azure vs Google Cloud
PostgreSQL in the cloud
- PostgreSQL Hosting: Best Providers & How to Choose
- Managed PostgreSQL Compared: RDS vs Cloud SQL vs Azure vs Neon
- Free PostgreSQL Hosting: Best Options & Their Limits
- Serverless PostgreSQL: Neon, Supabase, Aurora Serverless & More
- Neon Postgres Review 2026: Features, Pricing & Limits
- Supabase Review 2026: Postgres, Auth, Pricing & Limits
- Cloud PostgreSQL: Options on AWS, Azure & Google Cloud
AWS
- What Is AWS RDS? Amazon RDS Explained (Engines, Pricing, Limits)
- AWS RDS Pricing Explained (With Worked Examples)
- AWS RDS for PostgreSQL: Setup, Pricing & Best Practices
- AWS RDS for MySQL: Setup, Pricing & Best Practices
- AWS RDS for SQL Server: Editions, Licensing & Setup
- AWS RDS for Oracle: Licensing, Pricing & Setup
- Amazon Aurora Explained: Architecture, Pricing & When to Use It
- Aurora PostgreSQL: Setup, Pricing & Performance Tips
- AWS RDS Free Tier: What's Included & How to Avoid Charges
- Amazon RDS Proxy Explained: When You Need It
- AWS Database Services Compared: RDS, Aurora, DynamoDB & More
- AWS RDS Multi-AZ vs Read Replicas Explained
- AWS RDS Backups & Snapshots: Restore Step by Step
- AWS Database Migration Service (DMS): Step-by-Step Guide
Microsoft Azure
- Azure SQL Database Explained: Tiers, Features & When to Use It
- Azure SQL Pricing Explained: DTU vs vCore, Serverless & Worked Examples
- Azure SQL Managed Instance: Features, Pricing & Migration
- Azure Database for PostgreSQL (Flexible Server) Guide
- Azure Database for MySQL Guide: Setup & Pricing
- Azure Database Services Compared: SQL, Cosmos DB, PostgreSQL, MySQL
- Azure SQL Database Free Offer: Limits & How to Use It
- Azure Cosmos DB Explained: APIs, Pricing & Use Cases
Google Cloud
- Google Cloud SQL Explained: Engines, Pricing & Setup
- Google Cloud SQL Pricing Explained (With Examples)
- Cloud SQL for PostgreSQL: Setup, Pricing & Best Practices
- Cloud SQL for MySQL: Setup, Pricing & Best Practices
- Google Cloud Database Services Compared
- AlloyDB Explained: Google's PostgreSQL-Compatible Database
- Google Cloud Spanner Explained: When You Need It
MySQL in the cloud
SQL Server in the cloud
MongoDB in the cloud
Redis in the cloud
Frequently asked questions
What is a cloud based database?
It is a database hosted on a cloud provider's infrastructure and accessed over the network. Most cloud databases are offered as managed services, where the provider runs the servers, operating system and database software, and handles backups and patching.
What is the best cloud based database?
There is no single best one. For a relational application, a managed PostgreSQL or MySQL service in the same cloud as your application is the usual starting point. Choose NoSQL only when your access pattern fits it, and a warehouse only for analytics.
Is there a free cloud based database?
Yes. Azure SQL Database, Neon, Supabase, MongoDB Atlas, Cloudflare D1 and others document free tiers, and AWS and Google Cloud offer free plans or trials with credits. Limits on storage, compute and idle time apply, and some are time-limited trials rather than always-free tiers.
What are AWS cloud database services?
The main ones are Amazon RDS (managed PostgreSQL, MySQL, MariaDB, SQL Server, Oracle and Db2), Amazon Aurora (including Aurora Serverless), Amazon DynamoDB for NoSQL and Amazon Redshift for analytics.
What is a managed relational cloud database service?
It is a service that runs a relational engine such as PostgreSQL, MySQL or SQL Server for you. Amazon RDS, Azure SQL Database, Azure Database for PostgreSQL and Google Cloud SQL are examples: you choose the engine and size, and the provider handles installation, patching, backups and failover.
Is a cloud database secure?
The providers secure the infrastructure, but you configure access: network rules, users and roles, encryption settings and who can reach the database from the internet. A cloud database is as secure as that configuration.
Sources
- AWS docs: What is Amazon RDS?
- Amazon RDS product page
- Amazon Aurora Serverless
- Amazon DynamoDB
- Amazon RDS Free Tier
- Microsoft Learn: What is Azure SQL Database?
- Microsoft Learn: Azure SQL Database serverless
- Microsoft Learn: Azure SQL Database free offer
- Microsoft Learn: What is Azure Cosmos DB?
- Google Cloud: Cloud SQL overview
- Google Cloud: Cloud SQL free trial instance
- Google Cloud: BigQuery pricing
- Neon docs: Scale to zero
- Supabase pricing
- Google Cloud: Firestore
- Google Cloud databases
- MongoDB Atlas pricing
- PlanetScale pricing
- Aiven pricing
Checked 8 October 2026.
How we research cloud database guides: our editorial method. CodeWithSQL earns nothing from the providers mentioned.