OpenSSL Communities

OpenSSL new project: Jostle

SK Shubham Kumar Tue 11 Nov 2025 12:36AM Public Seen by 29

Hi everyone, at the recent OpenSSL Conference held in Prague, I got to know about the new OpenSSL project, Jostle (https://github.com/openssl-projects/openssl-jostle) - a Java-based provider that brings the OpenSSL native cryptographic library into the Java ecosystem. I thought to give it a try and deployed it in a VM. I have also recorded the complete deployment and steps and would like to share those resources with everyone who wants to give it a try:

YouTube: https://www.youtube.com/watch?v=VjXnHwPAvic

Github: https://github.com/ngkore/jostle

NT

Nicola Tuveri Tue 11 Nov 2025 6:05AM

Thanks @Shubham Kumar ! It looks interesting! I think it would be interesting also to see "Jostle in action" after building and testing it: how do you plan to use Jostle in your research? can we see an example of a Java application using Jostle for its crypto backend?

SK

Shubham Kumar Thu 20 Nov 2025 12:52PM

Apologies for the late follow-up; I didn't receive any email. Sure @Nicola Tuveri , actually, I am planning to use Jostle to see its performance against Bouncy Castle's implementation, but I was a little busy lately, so I will keep you updated here once I get any updates.

SK

Shubham Kumar Sat 7 Feb 2026 1:25PM

Hi, everyone, sorry it took a while but finally I have completed the performance benchmark between Jostle and Bouncy Castle and prepared a series of documents for anyone to follow and deploy locally. And also tried to cover the main programs through which I am able to achieve this progress. Please check it out here - https://docs.ngkorefoundation.org/pqc/jostle-vs-bc/.

This is a JMH benchmark of commonly supported algorithms between these two java providers for fair comparison. I have also included a visualizer (https://ngkore.github.io/jostle-bc-crypto-benchmarking/) to explore key JMH metrics such as throughput, average time, sample time, and single-shot time, making results easier to interpret. It took me around 7-8 hours to generate the benchmarking results even though they have so much less in common. However, I believe the accuracy of the measurement can be increased greater with better parameters (but that will take much longer than this).

And I want to admit that initially I thought the Bouncy castle will show better results as it is a native java lib, but I was amazed to see jostle which is a JNI bridge to the native OpenSSL library has better performance that bouncy castle in most of the cases.