For years, the majority of internet applications have used either TCP or UDP as the transport layer protocol of choice. Recently, a new contestant - QUIC, an end-to-end encrypted protocol - has entered the ring. QUIC is being marketed as the successor to TCP and tackles some of its inherent problems. An excellent example is the Head-Of-Line (HOL) blocking issue perceived by, for example, web pages. While QUIC holds many advantages over its predecessor, it still suffers from interoperability challenges and holds a number of open questions. The biggest one is: how do we efficiently test, measure, and debug the performance of applications using QUIC?
Answering the above question is non-trivial and plays a significant part in my research as a Ph.D. candidate. In this talk, we will – from a beginner's perspective – discuss why getting started with optimizing applications using QUIC as a protocol might seem daunting. We will discuss some tools available such as qlog, and why they can be more insightful than solely relying on packet traces. Finally, we will demonstrate our in-house developed automation framework called Vegvisir. Initially designed for analyzing the performance of video streaming over QUIC, it is now generalized for end-to-end testing of any networked applications. The goal of Vegvisir is to help alleviate some of the obstacles mentioned earlier, but also to ensure that developers and researchers can perform tests with QUIC consistently and reliably.