Working with large data structures in memory poses certain restrictions on performance and scalability. These issues are especially important in Fintech as effective handling of large data sets and efficiency are integral parts of high-performance data processing, monitoring trading performance, risk management and low-latency analytics.

This presentation explores the differences of working with domain-oriented objects vs tabular data structures, where we will be demonstrating how similar tasks can be solved using standard libraries, new approaches, rich data abstractions and library-based optimizations that have been proven in production environments. These libraries represent the current state of the art in addressing the dual needs of performance and advanced data processing capabilities. 

We will focus on three library solutions for managing data based on an example of high-performace CSV processing, comparing their efficiencies from a variety of perspectives.