Research of Seismic Streaming Storage Based on Druid

Along with the growth of seismic observation networks, the amount of seismic observation data is also increasing rapidly with the form of fast arrival stream, and it is an important and challenging task to store seismic data in time. However, two bottlenecks (slow read/write speed, poor user concurrency) still make the practical applications difficult. We proposes an Apache Druid-based storage solution for seismic observation data streams. The solution is more adaptable to the read/write requirement of large-scale seismic data streams. Specifically, we achieve real-time observation for two aspects: (I) using Kafka as a distributed message exchange system; and (II) implementing accesses and exchange with massive seismic stream data onto the Kafka indexing service provided by Druid. The experimental results show that the Druid-based temporal data storage solution has better stream input and real-time query performance compared with the existed Hbase storage method.