POI SS UserModel Performance

Question

When using SS UserModel to parse xls files, whats the difference in performance compared to HSSF?

Answer 1

For .xls files, there is no performance difference between the HSSF UserModel and SS UserModel. That's because the SS UserModel for HSSF is largely just some very clever retrofitted interfaces around the existing code.

The performance with .xlsx files (either direct calls to XSSF, or transparently through SS) is slower and more memory intensive than HSSF. The .xls file format is a binary one, .xlsx is an XML based one (a zip of XML files). Writing the code to process the XML is simpler, which is a good thing! However, the overhead of the xml parsing means that more memory is needed, and it's slightly slower.