Aggregating billions of rows per second with SIMD

Tancrede Collard

Tancrede Collard

QuestDB Team

SIMD instructions are specific CPU instruction sets for arithmetic calculations that use synthetic parallelization. This approach allows us to perform the same calculations and operations on numerous data points simultaneously. This post describes how SIMD works with typical operation performance and describes additional optimizations we managed to achieve.

Speeding up InfluxDB line protocol

Tancrede Collard

Tancrede Collard

QuestDB Team

InfluxDB is the current market leader in time series. This post examines their ingestion format called InfluxDB line protocol (ILP) and compares data ingestion performance between QuestDB and InfluxDB. We'll look at data loss over UDP and some of the reasons why QuestDB is more efficient at ingesting records in ILP.