Amazon Web Services Inc. is beefing up its database services in a big way.

The company on Wednesday introduced a new offering at its AWS re:Invent conference that aims to boost the reliability, scale and usability of its Aurora database. In addition, it announced a brand-new service called AWS Neptune to help customers build and run graph databases.

With the announcements, Amazon is hitting back at its main rival in the database world, Oracle Corp., which went on the offensive last October with the launch of its “self-driving” 18c Autonomous Database service, claiming it costs around 50 percent less to run than Amazon’s alternatives.

Amazon’s answer to Oracle’s new database is the introduction of several new capabilities addressing a number of different use cases.

First up is the introduction of Aurora Serverless, based on the firm’s original MySQL and PostgreSQL-compatible relational database Aurora. With Aurora Serverless, users now have the option to create database instances only when they’re needed, which can be scaled up or down according to demand. The database shuts down when it’s not in use, giving customers the performance they need only when they need it, without having to worry about provisioning infrastructure beforehand.

Amazon said Aurora Serverless, available in preview, is priced by the second, which means customers won’t be paying for idle database services left running overnight.

Serverless for Aurora is not a surprise, but a consequence of AWS’s overall push for serverless,” said Holger Mueller, principal analyst with Constellation Research Inc. “Making it easier for developers to use the flagship AWS database is a good move.”

Aurora itself has been updated with a new feature called AWS Aurora Multi-Master, which is designed to help customers eliminate downtime. Aurora Multi-Master can scale out across multiple data centers and can distribute writes across multiple availability zones, creating tolerance for any failures. So, for example, if a user’s writes fail in one availability zone, it will not affect the performance of the application it powers. Aurora Multi-Master also provides faster write performance, the company said.

Amazon said Aurora Multi-Master is currently available in preview for single region distribution, but should work across all of its global regions by early next year.

The company also lifted the lid off of a new service called AWS Neptune, which is billed as a “fully managed graph database service.” Graph databases are increasingly popular with developers because they provide unique capabilities for building things such as social networks. For example, they can be used to map relationships between different entities within a network. However, graph databases are notoriously difficult to set up, Amazon said.

That should change with Neptune, which is a “purpose-built, high-performance graph database engine optimized for storing billions of relationships and querying the graph with milliseconds of latency,” said Amazon technical evangelist Randall Hunt.

Neptune also boasts some impressive failover features with data replicated in six locations across three availability zones, with automatic backups saved to the AWS Simple Storage Service.

“Neptune fill a key gap in the AWS DB portfolio — graph databases – which recently have seen a revival due to their ability to map relationships more efficiently,” said Constellation’s Mueller. “One more step from AWS to make themselves the one stop for all database needs in the cloud.”

Lastly, Amazon added some new features to its DynamoDB NoSQL key-value store service, including something called Global Tables and a new Backup and Restore feature. Global Tables expands on DynamoDB’s ability to replicate tables across availability zones, extending the feature to its global regions as well. The feature is designed for customers such as Expedia, which want the same kind of performance and low latency for their applications no matter in which part of the world they’re being used.

As for the new backup feature, this allows customers to set up continuous or on-demand backups for data stored within the service. The idea is to provide restore capabilities that can recover lost data in milliseconds.

Image: AWS/YouTube