Autonomous Databases: Self-Driving, Self-Securing, and Self-Repairing
A self-driving database is called an Autonomous Database. The most notable example of one is Oracle Autonomous Database. It is a fully managed database service that runs in the cloud. What sets an autonomous database apart is that it does not require an administrator to manage any aspect of the database. Typically, non-autonomous cloud-managed databases such as AWS RDS require an administrator to perform at least some of the tasks such as making backups.
An autonomous database eliminates the need for data management by a database administrator (DBA) which is error-prone. Patching, upgrades, tuning, scaling, and other routine database management tasks are fully automated and are performed while the database is running. The autonomous database offers several benefits and unmatched performance.
Autonomous—The autonomous database is self-driving, self-securing, and self-repairing. Self-driving implies that all routine administration tasks including provisioning, performing backups and recovery, making fault-tolerant failovers, applying patches, performing upgrades, optimizing and tuning for performance, and scaling are all fully automated. Self-securing implies that all data is encrypted and security updates & patches are applied automatically both periodically and off-cycle. The option for encryption keys to be managed by a user is typically provided. Self-repairing implies that the database detects and recovers from failure of any sort, such as a single node failing, without user intervention.
Machine Learning and AI—Machine Learning, AI, and automation are used to provide reliability, security, operational efficiency, and elastic data management; making an autonomous database highly performant. An autonomous database runs OLTP and analytic workloads up to 100x faster than regular databases. An autonomous database tunes itself using Machine Learning algorithms. It makes use of automatic indexes to accelerate applications.
Advanced Features—Before applying patches and upgrades the autonomous database replays the production workload on a test database to verify that an upgrade or patch does not incur any unexpected side effects on a mission-critical system.
Highly Available—An autonomous database is designed for mission-critical applications and guarantees high (99.995%) uptime. Maintenance patching, upgrades, and repair are performed in a rolling fashion across the nodes in a cluster without incurring any downtime.
Auto Scaling—Compute and Storage are scaled elastically as needed with no downtime.
Fault Tolerant—The database recovers automatically from failure with high ( 99.995%) uptime guaranteed.
Types of Autonomous Database—An autonomous database can be Autonomous Transaction Processing (ATP) or Autonomous Data Warehouse (ADW). Each of these are designed for a different type of workload. ATP is optimized for online transaction processing (OLTP) and mixed workloads. ADW is optimized for data warehousing workloads.
Dedicated Cloud Infrastructure Option—An autonomous database dedicated service is typically available to run on dedicated cloud infrastructure. The dedicated service provides an isolated private database cloud. The dedicated cloud infrastructure option provides the highest level of workload isolation, security, reliability, and customizable operational policies.
