Microsoft Azure - DataStax partnership focuses on hybrid cloud and IoT applications

Cassandra's masterless architecture gives it a unique advantage in developing multi-data-centre hybrid apps, claims CEO Billy Bosworth

DataStax, the company that delivers the Apache Cassandra NoSQL database as the DataStax Enterprise package (DSE), today announced a partnership with Microsoft to deliver DSE on the Azure cloud platform.

Already available on Amazon Web Services (AWS), where it supports the likes of media-streaming service Netflix, DataStax CEO Billy Bosworth (pictured) described the new relationship with Microsoft as an "magical partnership".

"We have alignment at the executive level, the technical level and the field level," he told Computing.

DataStax says DSE For Azure is the "first and only fully distributed database platform to run on public, private and hybrid cloud environments".

This claim is justified, according to Bosworth, because of Cassandra's masterless peer-to-peer architecture.

"There is no concept of a master or primary, and that is the true definition of a 'fully distributed database'," he said. "Lots of databases can be distributed horizontally, but fully distributed means an architecture with no single points of failure and no control points."

DynamoDB (on which Cassandra is based) has a similar architecture but is proprietary to Amazon, he went on, whereas DSE can run on AWS, Google and now Azure - and also in customers' own data centres.

Hybrid cloud has become possible because of recent advances in data centre technology, open APIs, and the maturity of cloud platforms, particularly IaaS, he said, but databases have held back progress.

"There has been one anachronism in the whole scheme: the database," Bosworth said. "If you take a master-slave architecture and try to make it work in a distributed environment you find you have an anchor around your neck. A fully distributed architecture allows developers to leverage multi-data-centre systems and benefit from them rather than having to work around them."

This peer-to-peer architecture reduces the complexity of developing distributed applications that need to run in different environments, as developers don't need to take into account where the control systems are to be located, said Bosworth.

Asked about the target market for DSE on Azure, Bosworth mentioned firms running multiple data centres where web or IoT applications must be always available, cope with very high transactional volumes and deliver extremely low latency.

"The only way to solve these three problems is to have a fully distributed system that can move the data out closer to the endpoints geographically," Bosworth said, describing DSE's main use case as supporting "contextual applications in the moment".

Such applications typically involve personalisation, for example, recommendation engines, playlist management and session management across multiple devices. Related to this are anti-fraud solutions, which flag up deviations from expected usage patterns. Others use DSE as a distributed object store to support product catalogues with real-time searching.

Other NoSQL databases available on Azure include Apache HBase, MongoDB, Couchbase, Cloudant, DocumentDB and HDInsight.