SnappyData offers developers and users a way to do simplified, agile analytics on data in motion and at rest, explained SnappyData chief technology officer Jags Ramnarayan, with application areas including IoT, including financial services, manufacturing, and telecommunications, according to the company.
It uses Apache Spark, and therefore can be connected to systems from most vendors. SnappyData can be used to analyse live streams of data, operational data, or big historical data, Ramnarayan said.
Recently acquired by Tibco, SnappyData was spun out from Pivotal in 2016.
- Using it as an in-memory data store for Tibco Spotfire and Tibco Data Science.
- Building the "ultimate digital business analytics platform for streaming IoT data" by combining Spotfire, Data Science, SnappyData and Tibco Streaming Analytics.
- Broadening the company's commitment to open source.
By fusing Spark with an in-memory database, SnappyData provides very fast reads and writes (it is capable of ingesting millions of records per second, according to Ramnarayan), takes advantage of multi-core CPUs by providing a row/column in-memory store, and allows Spark streams and mutable tables to be joined in a single query.
Advantages include up to 20 times the speed of Spark's native caching (thanks to better memory management and vectorisation), up to 100 times the speed of Spark with a conventional database, dramatic performance improvements by using probabilistic data.
Other application areas mentioned by Ramnarayan included predictive maintenance, smart cities, media (the delivery of targeted advertising is "a complex thing," he observed), and security.
All of these involved the ingestion of live data, he pointed out.
Disclosure: The writer attended Tibco Now 2019 as a guest of the company