Case Study: Coralogix achieves 349x faster query processing with ScyllaDB

How Coralogix cut processing times from 30 seconds to 86 milliseconds with a PostgreSQL to ScyllaDB migration

Coralogix, an observability platform, needed a faster way to process Parquet metadata for its distributed query engine. Its original metastore, built on PostgreSQL, couldn’t deliver the low-latency, high-scale performance required to support heavy query and ingest workloads across large volumes of archived data.

Coralogix migrated the metastore to ScyllaDB, using a carefully redesigned data model for blocks, bloom filters, and column metadata. With ScyllaDB, the team cut query processing time from 30 seconds to 86 milliseconds, handled around 10K writes per second with P99 latency under 1 ms, and reduced 50K-file processing to under 500 milliseconds.

Coralogix

Dan Harris

Principal Software Engineer