Table of Contents
Fetching ...

From Files to Streams: Revisiting Web History and Exploring Potentials for Future Prospects

Lucas Vogel, Thomas Springer, Matthias Wählisch

TL;DR

The paper investigates how historic HTTP design choices and evolving web content practices have rarely been aligned, limiting latency reductions. It analyzes HTTP evolution from 0.9 through 3 and traces how content creation shifted toward dynamic, bundled assets, contributing to render-blocking and larger pages. The authors propose fine-grained content segmentation to exploit streaming capabilities in HTTP/2 and HTTP/3, aiming to render content progressively without protocol changes. They identify three core challenges—detecting content usage, ordering and interleaving pieces, and streaming delivery—and argue that addressing these could substantially speed up page rendering in real networks.

Abstract

Over the last 30 years, the World Wide Web has changed significantly. In this paper, we argue that common practices to prepare web pages for delivery conflict with many efforts to present content with minimal latency, one fundamental goal that pushed changes in the WWW. To bolster our arguments, we revisit reasons that led to changes of HTTP and compare them systematically with techniques to prepare web pages. We found that the structure of many web pages leverages features of HTTP/1.1 but hinders the use of recent HTTP features to present content quickly. To improve the situation in the future, we propose fine-grained content segmentation. This would allow to exploit streaming capabilities of recent HTTP versions and to render content as quickly as possible without changing underlying protocols or web browsers.

From Files to Streams: Revisiting Web History and Exploring Potentials for Future Prospects

TL;DR

The paper investigates how historic HTTP design choices and evolving web content practices have rarely been aligned, limiting latency reductions. It analyzes HTTP evolution from 0.9 through 3 and traces how content creation shifted toward dynamic, bundled assets, contributing to render-blocking and larger pages. The authors propose fine-grained content segmentation to exploit streaming capabilities in HTTP/2 and HTTP/3, aiming to render content progressively without protocol changes. They identify three core challenges—detecting content usage, ordering and interleaving pieces, and streaming delivery—and argue that addressing these could substantially speed up page rendering in real networks.

Abstract

Over the last 30 years, the World Wide Web has changed significantly. In this paper, we argue that common practices to prepare web pages for delivery conflict with many efforts to present content with minimal latency, one fundamental goal that pushed changes in the WWW. To bolster our arguments, we revisit reasons that led to changes of HTTP and compare them systematically with techniques to prepare web pages. We found that the structure of many web pages leverages features of HTTP/1.1 but hinders the use of recent HTTP features to present content quickly. To improve the situation in the future, we propose fine-grained content segmentation. This would allow to exploit streaming capabilities of recent HTTP versions and to render content as quickly as possible without changing underlying protocols or web browsers.
Paper Structure (19 sections, 1 figure)

This paper contains 19 sections, 1 figure.

Figures (1)

  • Figure 1: Historical comparison of Web technologies, including protocols and content creation and presentation features such as bundlers, JavaScript, and CSS. A major change occurred when HTTP/2 was released (dotted line). Protocol deprecations are not shown. Source for asset distribution data: HTTP Archive httpArchivePageWeight. Images sizes are not stacked, as images are not render-blocking.