The post discusses a case study from Aladdin Wealth Tech, detailing how they identified and fixed a performance issue with a Java hash table handling very large portfolios in an Apache Spark job. It was discovered that hash collisions in the Apache Commons Lang3 Pair class were causing quadratic time complexity (O(N2)). By restructuring their map keys and avoiding the use of problematic hash functions, they restored the expected performance. This experience underscores the importance of understanding data structures and their underlying mechanics to prevent such issues.
Table of contents
Solving the Mystery of the Slow Hash TableVery Large PortfoliosJMH BenchmarkPosition ContributorsHash Table BasicsHash CollisionThe FixLessons learned35 Impressions