I've tested the performance of HTTP/2 (H/2) and HTTP/1.1. I hosted a Rails app with different numbers of requests on various pages:
- The landing page had approximately 25 requests for resources like JavaScript, CSS, and images.
- Another page had around 50 requests in total.
- A different page had about 100 requests, primarily for images.
- A page had around 500 requests, primarily for images.
- A page with approximately 1000 requests for images.
I hosted this app on AWS ECS and added two load balancers: one with HTTP/2 support and the other with HTTP/1.1 support. Then, I included these endpoints in the Page Speed test of Pingdom. Here are the results.
As you can see here, performance difference is there when there more than 100 requests in a page. But for less than 100 requests, HTTP/1.1 gave better result than HTTP/2.0. Is there any reason for this? or Should I try to test with different approach?
