Service Models

What Is Database as a Service (DBaaS)?

What database as a service (DBaaS) is, what the provider manages versus what you still own, and how to choose between relational and NoSQL options.

On this page
  1. Short answer
  2. What the provider manages, and what you still do
  3. Is DBaaS a PaaS?
  4. Relational and NoSQL options
  5. Availability and backups
  6. Serverless and billing
  7. Lock-in and portability
  8. Common mistakes
  9. Security basics
  10. Frequently asked questions

Short answer

Database as a service (DBaaS) is a cloud service in which the provider runs the database software and the infrastructure under it. The provider typically handles provisioning, patching, backups, replication and failover. You still own the schema, the queries, the access rules and the data. DBaaS removes much of the operational work, but it does not remove the need for database design and care.

What the provider manages, and what you still do

Usually handled by the provider Still your responsibility
Provisioning and setup Schema design
Operating system and engine patching Indexes and query tuning
Automated backups and point-in-time restore Users, roles and access control
Replication and failover The data itself, including retention
Scaling of capacity Application logic and connection settings

The exact split varies by provider and service, so read the shared responsibility documentation for the one you use.

Is DBaaS a PaaS?

Many sources count DBaaS as a type of platform as a service (PaaS). You usually get no operating system or SSH access. Amazon RDS Custom is one exception, because it offers more access to the underlying environment.

Some vendors argue that DBaaS is not PaaS, because it sits at the application layer on top of the platform. The label matters less than the responsibility split in the table above. For the wider service models, see IaaS vs PaaS vs SaaS compared, and for how services are classified, see IaaS or PaaS? Azure Storage, App Engine and Elastic Beanstalk.

Relational and NoSQL options

Most DBaaS offerings are relational, but NoSQL options are common too. Some examples:

Type Examples Typical reason to choose
Relational Amazon RDS, Azure SQL Database, Google Cloud SQL, Cloud Spanner, Neon, PlanetScale You need SQL, joins and transactions
NoSQL Amazon DynamoDB, Azure Cosmos DB, MongoDB Atlas, Firestore You need flexible data models or key-based access at scale

Amazon RDS launched in October 2009, starting with MySQL. It now supports several engines, including MySQL, PostgreSQL, MariaDB, Oracle, SQL Server, Db2 and Amazon Aurora. Choose the engine by your application’s needs and by which features the service supports.

Availability and backups

Managed services usually offer high-availability options. In Amazon RDS, Multi-AZ deployment keeps a synchronous standby copy in another availability zone. Failover typically takes one to two minutes, though this varies. RDS Multi-AZ carries a 99.95% monthly SLA. DynamoDB carries 99.99%, and 99.999% for global tables.

Automated backups are useful, but they are not a disaster recovery plan on their own. RDS keeps automated backups for 0 to 35 days, and setting retention to 0 turns them off. Within that window you can restore to a point in time. Test restores regularly, and keep copies in another region if you need protection against a regional outage.

Serverless and billing

Some services scale capacity with demand. Aurora Serverless v2 and DynamoDB on-demand are examples, and Neon can scale to zero when idle. Serverless options can reduce idle cost, but check how they bill and how quickly they scale for your workload.

DBaaS billing usually combines several items: instance hours or capacity units, storage per GB-month, I/O requests, extra backup storage and data transfer. A quick estimate that covers only instance size can miss a large part of the bill.

Lock-in and portability

Engines that match open-source databases, such as MySQL or PostgreSQL, usually make migration easier. Proprietary engines such as DynamoDB, Google Cloud Spanner and Azure Cosmos DB can make it harder to move, because their APIs and features are specific to the provider. Lock-in is a trade-off, not an automatic problem, so weigh it against the benefits you get.

Common mistakes

  • “DBaaS will tune my queries.” It will not. Slow queries and missing indexes are still your job.
  • “Backups mean I am protected.” Backups help with some failures, but you should test restores and plan for regional outages.
  • “DBaaS means NoSQL.” Most DBaaS offerings are relational.
  • “A managed service is always secure by default.” Access settings are still yours to set and review.

Security basics

Do not make database instances publicly accessible. Keep them in private subnets, or allow only specific IP addresses. Use accounts with the least privilege needed, review who can connect, and rotate credentials when people or systems change.

Frequently asked questions

Does DBaaS mean no database administration?

No. The provider takes on much of the routine work, such as patching, backups and failover. You still design the schema, tune queries, control access and decide how long to keep data.

Is DBaaS the same as Data as a Service?

No. Data as a Service (DaaS) is about turning data into useful information, and it is not a technology service in the same sense. DBaaS is about running database engines that store and serve data, along with the infrastructure they need.

For a related managed service that is often classed as a subset of PaaS, see what backend as a service is.

The complete guide to service models

IaaS vs PaaS vs SaaS: Cloud Service Models Compared With Examples
Read the guide

Keep exploring

Guides from other topics

All topics →