Michael Mifsud writes about how HTTP/2 might not be a magic bullet for performance improvement. He relates his particular use case over at 99designs for image-heavy pages.
His post makes for an interesting engineering study, but I think it fails to put all of the benefits of HTTP/2 into perspective, instead focusing on somewhat minor time-to-view measurement differences (5-10%, not to belittle them, isn’t that drastic, and is specific to his use case and setup).
Very few new technologies are 100% wins for everyone, but despite this analysis, I think it’s pretty clear HTTP/2 has a lot more going for it than against it.
