Case Study: Natura Achieves Beautiful Results with ScyllaDB

Natura cuts batch processing from 6 hours to under 10 minutes with ScyllaDB

Natura, a major Brazilian beauty products manufacturer, faced significant performance and cost challenges with its Apache Cassandra database. The system, part of a Kafka and Spark streaming architecture handling 400,000 messages daily, suffered from high latency due to Java garbage collection and was a hardware hog, requiring an expensive 5-node cluster with 11 primary disks.

By implementing ScyllaDB, Natura achieved a dramatic improvement. The switch resulted in a 10% reduction in batch processes, with some jobs dropping from 6 hours to under 10 minutes. Write latencies improved from milliseconds to microseconds, and hardware needs were reduced to just 2 primary disks, cutting costs by over 50%. ScyllaDB provided superior performance, significant cost savings, and eliminated their issues with the JVM.

Natura