SQream’s GPU-driven data warehouse screams faster

The v3.0 version of SQream DB loads data warehouses up to 40 per cent faster and can boost query speed 15X.

The upgrade makes better use of Nvidia Tesla Core 100 GPUs with the code tweaked for up to 15x faster queries for multi-table joins and count distinct operations compared to previous versions of SQream.

Other improvements include:

  • Data stores can be ingested and analyzed faster with compressed Parquet files 
  • External Table enables direct access to data for fastest available analysis
  • Optimised Spark connector optimized for two-way interconnect, using  SQream DB’s native communication protocol, enables faster integration into data science pipelines
  • Dynamic Workload Management provides on-the-fly resource workflow prioritisation
  • Reduced complexity with built-in auto-compress and SQL query optimiser
  • Easy to deploy Docker container images makes for rapid and efficient implementation

SQream says its External Table syntax  offers additional flexibility compared with traditional flat-file bulk loads. It has a Dynamic Workload Management facility to prioritise projects and allow on-the-fly changes to resource allocation.

The legacy on-premises data warehouse world have a fight on their hands. They need to match SQream speed and take public cloud hosting of data warehouses on board with hybrid product capability.

And they face competition from Snowflake Computing’s data warehouse in the public cloud and from SQream’s GPU-driven acceleration to produce faster query results for data warehouse users. SQream can run either on-premises or in the public cloud to complicate matters.