• Ips4o
• dbms
• sorting
• algorithm
• algorithm selection machine learning

In computer science, sorting is vital for structuring and efficiently retrieving data. Its role extends beyond listing numbers or alphabetizing words; it's critical in optimizing algorithms, databases, and system performance. Sorting, in essence, is integral to the functionality of our digital ecosystems, from financial systems to social networks.

We partnered with a Database Management System (DBMS) company to understand the unique sorting challenges within their environment, examining how these deviate from conventional academic and theoretical evaluations. We dove into both widely-used and cutting-edge sorting algorithms, exploring their application in a DBMS context and their respective trade-offs.

Following this, we devised a benchmark suite to rigorously analyze these algorithms, shedding light on their strengths and weaknesses. Our research culminates in a machine learning approach for selecting the most efficient algorithm based on dataset characteristics.