20 years have passed since the IETF introduced HTTP/1.1. At the time web pages were still designed with tables and font tags and they had hardly more than four images per page in total. HTTP/1.1 was perfectly suited to this application scenario. For today's projects, however, HTTP/1.1 is a massive bottleneck. To compensate for its conceptual disadvantages, we developed build processes in which we bundle resources via concatenation, inlining or spriting. In addition, we rely on concepts such as domain sharing and cookieless domains. The good news is: HTTP/2 has arrived and puts an end to the disadvantages of its predecessor. And it is supported by all modern browsers. But it's not enough to just flip a switch. We need to rethink and throw established processes and concepts overboard if we are to get the most out of HTTP/2.