Aerospike Gives Legacy Infrastructure a Real-Time Boost
A database connector upgrade released this week by Aerospike Inc. links open source frameworks like Apache Spark data streaming to existing enterprise data infrastructure.
Among the goals is providing backward compatibility with mainframes and other relational data stores that are currently unable to handle data analytics and machine learning workloads. Rather than replacing those assets, Aerospike’s connector makes its Database 5 platform plug-compatible with enterprise data fabrics.
Along with Spark, the real-time NoSQL database vendor said Tuesday (Sept. 15) its plug-compatible connector integrates Database 5 with Apache Kafka, Pulsar and Java Messaging Service. Those links could, for example, allow mission-critical data architectures to scan hundreds of terabytes of data using real-time Spark streaming.
The database connector upgrade would allow mainframes and other data infrastructure to handle real-time data workloads used to train machine learning models, according to Srini Srinivasan, Aerospike’s chief product officer and founder.
The bi-directional database connector could help reduce data scans from half a day to about 2.5 hours, Srinivasan added in an interview, enabling real-time database access and reduced latency.
The connector “augments traditional systems to create real-time data-driven systems that extend and enhance rather than rip and replace existing functionality” that are vital for capturing data and business processes, the company said.
Aerospike, Mountain View, Calif., said its Connect upgrade supports Spark streaming APIs that target real-time use cases like applying machine learning to fraud detection and customer engagement. Those and others increasingly require low latency.
Also supported is a connection of the Aerospike database to enterprise streaming pipelines based in Kafka. “The integration enables Aerospike Database 5 to pull data from Kafka and to submit Aerospike change notifications via the optimized Database 5 protocol to Kafka,” the company said.
Support for the Java Messaging Service allowing enterprises to use JMS stream data into and out of the Aerospike database. JMS is a preferred means of synchronizing with mainframe applications.
Apache Pulsar is also supported via an outbound connector that delivers change notifications from the Aerospike platform via a database protocol to the Pulsar distributed messaging and streaming platform. That feature targets streaming pipelines for applications ranging from the Internet of Things to log analysis.
The non-relational database vendor added Optane persistent memory to its flagship database beginning in late 2018. Srinivasan noted that its hybrid memory architecture that also includes DRAM boosts storage capacity on a single node.
For fintech applications, the new connector provides access to real-time data pipelines augmented by cached memory. Aerospike said a large, unidentified brokerage firm is using its connector to integrate real-time data access into its IBM mainframes.