Analysis of Next-Generation Internet Transport Protocols: QUIC, WebTransport, HTTP/3

Abstract

Introduction. Traditional Internet transport relies on the TCP/IP stack with applicationlevel protocols such as HTTP/1.1 and HTTP/2. This combination is reaching performance limits due to head-of-line blocking, multi-RTT handshakes, and the lack of built-in security at the transport layer. Modern latency-critical applications, including cloud gaming and AR/VR, require end-to-end latencies well below 50 ms, often below 20 ms, which the classic TCP+HTTP/1.1/2 stack struggles to provide. New transport solutions based on QUIC, combined with HTTP/3 and the WebTransport API, are designed to overcome these inefficiencies while preserving the Web programming model. Their global adoption has already grown significantly (over 40% of web traffic via QUIC/HTTP/3), which makes a systematic analysis of these protocols both timely and practically important.

Problem statement. Despite the rapid deployment of QUIC, HTTP/3, and WebTransport by major cloud providers and browsers, there is still a lack of consolidated analysis that connects IETF specifications, academic research, and real-world content delivery network (CDN) and 5G deployments. Practitioners often rely on scattered blog posts and partial benchmarks, which makes it difficult to understand where QUIC-based transports outperform the classic TCP+HTTP/2 stack, how different congestion control algorithms behave, and what limitations remain in latency-critical scenarios.

Purpose of this paper is to perform a critical analysis of next generation Internet transport based on QUIC, HTTP/3, and WebTransport, focusing on their architectural evolution, congestion control algorithms, deployment models, and security properties. To achieve this aim, the paper traces the evolution from SPDY and HTTP/2 to HTTP/3 over QUIC, compares congestion control schemes (CUBIC, BBRv2, and HyStart++ variants), reviews deployment optimizations (XDP and eBPF offload, 5G L4S), and identifies open issues such as multicast support, satellite links, and observability.

Methods. We survey IETF RFCs (for example RFC9000, RFC9114, and the QUICv2 draft) and recent research and industry reports. Performance is inferred from published benchmarks and experimental studies. The work has a survey and analytical character; original in house experiments are limited to indicative tests that complement these benchmarks. We also include anecdotal insights from our CDN and research experience (“under the hood” observations) to illustrate real world behavior.

Results. QUIC’s streamlined handshake (1-RTT, optional 0-RTT) and encrypted headers significantly reduce latency. Adoption is confirmed by major tech firms (Meta: ~75% traffic on QUIC/HTTP3). QUIC’s congestion controls show trade-offs: CUBIC is well-tested but can overshoot, whereas BBRv2 offers robust throughput in shallow buffers, with HyStart++ alleviating startup losses. WebTransport extends HTTP/3 with multiplexed streams and unreliable datagrams, now supported in browsers (e.g. Firefox 114). Offload strategies (XDP / eBPF) and libraries like Cloudflare’s quiche improve edge performance.

Conclusions. (1) QUIC/HTTP3 effectively address TCP/UDP bottlenecks, e.g. eliminating TCP’s head-of-line blocking. (2) Security enhancements (TLS1.3, header protection) mean almost entire packet payloads are encrypted. (3) Emerging CC algorithms like BBRv2 and hybrid slow-starts improve fairness and reduce loss. (4) WebTransport enables new web architectures (streaming, gaming) by combining QUIC with WebCodecs. (5) Practical deployment benefits from kernel-bypass (XDP) and QUIC libraries (quiche, lsquic). (6) Open issues include multicast-capable QUIC (e.g. MCQUIC), satellite scheduling, and richer logging (qlog) for observability. Future work may explore PERC-over-QUIC (forward-error correction), pluggable CC frameworks, and QUICv2-based enhancements.