Graph Databases Ascend to the Cloud
Bucking the trend toward open source databases, cloud vendors continue to promote proprietary graph databases that combine the ability to handle multiple data models with the distribution of data across different cloud regions.
While Amazon Web Services (NASDAQ: AMZN) has embraced an open-source approach with its new Neptune graph database, cloud database competitors Microsoft and Oracle (NYSE: ORCL) are betting there is plenty of life—and revenues—in cloud-based proprietary approaches. Oracle is promoting its upcoming “self-driving database” that leverages AI features to automate administrative tasks. Automation makes the database cheaper to run in the Oracle cloud, Oracle CTO Larry Ellison claimed last month.
Microsoft is meanwhile zeroing in on the vibrant graph market with a multi-modal graph database called Azure Cosmos DB.
Multi-modal graph databases are designed to support different model types such as a combination of document and key-value store along with a graph capability. Observers note that among the advantages of the approach are that graph and key value queries can be run against the same data. The downside is that performance cannot match a dedicated database management system, they add.
Along with Azure Cosmos DB, other multi-modal graph databases include ArangoDB and Sqrrl.
Microsoft (NASDAQ: MSFT) released Azure Cosmos DB details in December, including specs on APIs used to access and query data. Emphasizing the ability to distribute data across different Azure cloud regions, the company released a “multi-homing API” designed to reduce latency by identifying customers’ nearest cloud region, then sending data queries to the closest datacenter.
In support of its multi-model approach, Microsoft also released a batch of APIs that support SQL, MongoDB, Cassandra, Graph (Apache Gremlin) and a key-value database service called Table API. The company said in December that it plans to add support for other data models.
Microsoft is aiming Azure Cosmos DB at the emerging Internet of Things market along with web and mobile applications requiring “massive amounts of data, reads and writes at a global scale with near-real response times for a variety of data,” the company said.
The introduction Azure Cosmos drew mixed reviews in a lively, detailed discussion on the Azure Cosmos DB web site. After identifying several shortcomings, one long-time Azure storage table user concluded that the new database “is not production ready.”
Another user raised an issue that cloud vendors are likely to hear frequently in coming months: A globally distributed database may reduce latency, “but we need to give our users the choice of where their data is stored—particularly in regards to the [General Data Protection Regulation] that goes live in May 2018″ within the European Union.