Implementation of HP Vertica Increased BI system performance at NP Sovet Rynka.

Implementation of HP Vertica Increased BI system performance at NP Sovet Rynka

About project

iCore increased the performance of the monitoring system at NP Sovet Rynka by integrating it into a HP Vertica column-store DBMS.

Client НП «СОВЕТ РЫНКА»
Industries Energy
Solutions IT for Business

Extra

Project Aims and Objectives

  • Implement tools that allow quick processing of large data arrays coming from several dozen information systems.

Solutions Used in the Project

iCore proposed implementing HP Vertica column-store DBMS. Thanks to a lack of indices, advanced compression algorithms and distributed processing mechanisms used when analytic sections are generated, the platform has no limitations inherent to relational databases.

iCore carried out a field test implementation by deploying HP Vertica at the cluster with 3 servers and having loaded 1 TB data. Tests showed that the DBMS has high performance and allows easy operation with huge samples and generating unique reports based on any parameters.

During implementation, the customer’s IT infrastructure has been taken to a private cloud, built on the basis of VMware. HP Vertica nodes were placed on 3 virtual machines for automatic scaling if a load is increased.

The TalenD system is used as the ETL-tool. Integration with the basic Oracle DBMS and SAS Enterprise’s analytics tools is implemented using direct access drivers. To generate and visualize queries, Tableau Server and Tableau Desktop solutions are used, as well as the R analytical programming language.

Results and opportunities

The main benefit of HP Vertica implementation is expansion of tasks that can be completed using analytics tools. Processing long-term statistics allows the building of a richer mathematical market model, to discover hidden trends and more precisely predict price variation. Moreover, now it is possible to quickly build reports in any section, including those urgently requested by state regulators.