(AMZN -0.16%) is introducing a slew of new features for Amazon Web Services at its annual AWS re:Invent conference, which runs through Friday. The company is far and away the largest cloud infrastructure provider, and constantly adding new features and products is one way it plans to stay ahead of the competition.

One area that represents a major opportunity for AWS is the database market. Amazon is a small fish in a pond dominated by Oracle (ORCL 0.62%), but it can win market share as an increasing number of enterprises move from on-premise to the cloud. Amazon's database market share was just 2.3% in 2016, according to Gartner, putting it far behind Oracle's dominant 41.6%.

The Amazon Web Services logo.

Image source: Amazon.

New database products

Amazon announced two major new database products during re:Invent. Amazon Neptune is the company's entry into the managed graph database market. Unlike a relational database, which stores data in rows and columns much like a spreadsheet, a graph database stores data as nodes that can be linked together. This allows complicated hierarchical structures to be stored and retrieved efficiently. One possible application would be a social network, where relationships between users can be modeled as a graph.

The second new database product from Amazon won't be available until early 2018. Amazon Aurora Serverless, based on Amazon's Aurora relational database service, is designed for workloads that are highly variable. With the standard Aurora service, the user pays an hourly rate for a certain configuration as long as the database is running. But this isn't ideal for use cases where database access is intermittent and infrequent. An example might be a daily logging of data.

Aurora Serverless solves this problem by separating the underlying storage from the processing. Computer and memory are paid for by the second, and only when the database is actually being used to read or write data. Storage costs are separate, allowing a user to only pay for storage if the database is not in use.

There are other managed database products that use a similar pricing model. IBM's Cloudant, a nonrelational database service that stores data as documents rather than rows in a table, charges based on storage and throughput, for example. But there doesn't appear to be anything directly comparable to Aurora Serverless from any of the other major cloud infrastructure providers.

Database price wars

The Aurora Serverless announcement comes about two months after Oracle took a swing at AWS during its OpenWorld conference. Oracle announced the Autonomous Database Cloud, a database service that uses machine learning to automate upgrades, patches, and security updates, all while tuning the database on the fly.

The big news, though, was pricing. The details were left out, but Oracle executive chairman Larry Ellison said that Oracle would guarantee, in writing, that the Oracle autonomous database would cost at most one-half of what Amazon charges for its Redshift database service, which is a data warehousing product. Oracle's autonomous cloud will be available for data warehousing workloads by the end of the year, with other offerings coming later.

Oracle is aiming to not only retain its existing on-premise database customers looking to move to the cloud, but also win over cloud-first customers who would have otherwise chosen AWS or another cloud infrastructure provider. Meanwhile, Amazon is aiming to rip away customers from Oracle and other legacy database providers.

Oracle won't cede its market share without putting up a fight, but that may mean the company won't enjoy the same margins that it's used to as more customers move to its cloud database offerings. Amazon is still a minor database player, but that probably won't be the case indefinitely.