Does More Bandwidth Really Not Matter (Much)?

Seraj Al Mahmud Mostafa 1, Mike P. Wittie 1, Utkarsh Goel 2 · 2025

Read Does More Bandwidth Really Not Matter (Much)? by Seraj Al Mahmud Mostafa 1, Mike P. Wittie 1, Utkarsh Goel 2 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-03641.txt.

Introduction

Seraj Al Mahmud Mostafa, Mike Wittie, and Utkarsh Goel test the old claim that bandwidth barely matters for web speed, in this 2025 arXiv paper. They define a measure, the Critical Path of Improvement, that shows which network trait a given page would gain most from. Forty-five websites were scored and set against four mobile carriers in fifty-seven American cities.

Eighteen percent of the sites were mainly bandwidth-limited, and most others felt bandwidth at least partly. Latency still rules many pages, but the blanket dismissal of bandwidth no longer holds. The authors close with steps developers and carriers can take to spot and ease bandwidth crunches.

Catalog record

CREATORSeraj Al Mahmud Mostafa 1, Mike P. Wittie 1, Utkarsh Goel 2
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE27 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-03641

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Does More Bandwidth Really Not Matter (Much)? Seraj Al Mahmud Mostafa 1, Mike P. Wittie 1, Utkarsh Goel 2 Abstract The prevailing wisdom is that more network bandwidth does not matter much and that website performance is primarily limited by network latency. However, as mobile websites become more complex and mobile network performance improves, does this adage continue to hold? To understand the effects of small changes in network bandwidth and latency on website performance, we propose a novel webpage characterization metrics – Critical Path of Improvement (CPI). We compute CPI for 45 websites and analyze it against the network performance of four mobile ISPs in 57 US cities. Our results show that 18% of websites are primarily limited by bandwidth with others limited by bandwidth to some extent. These results show that contrary to accepted wisdom, insufficient bandwidth is a limiting factor in some website/network combinations. We also offer a discussion of approaches website developers and mobile network administrators can follow to understand and mitigate bandwidth limitations to website performance. † † institute: Montana State University, Bozeman, MT, USA and Akamai 1 Introduction The evolution of mobile web services goes hand-in-hand with the improvements in mobile network performance. As mobile networks offer lower latency and higher bandwidth, mobile websites may transfer more content to offer a richer experience, while maintaining good response times. But what are there instances where website design and network performance do not match up well? The prevailing wisdom is that network latency is the limiting factor to website performance. Two key articles driving this idea came from Mike Belshe Belshe2010More and Ilya Grigorik Grigorik2012Latency . They observe that beyond some point, adding more bandwidth results in diminishing marginal returns of website load time. Incremental reduction of latency, on the other hand, results in corresponding improvement of website performance. It would seem then that beyond a certain minimal bandwidth website performance is limited by network latency. While that may have been true a decade ago for landline internet and desktop websites, much of today’s web traffic involves mobile networks and smartphones. The mobile networks themselves have undergone significant improvement in the last several years. OpenSignal reports that from February, 2016 to January, 2020 4G latency (averaged across providers) improved from 75.5 ms to 52.9 ms (by 29.9%) and bandwidth from 9.7 Mbps to 25.6 Mbps (by 163%) 2020OpenSignal ; 2016OpenSignal . At the same time, there remain pronounced regional differences in mobile network performance. Specifically, in January, 2020 the variation of mean latency between US cities was as large as 37.8 ms and the variation of mean bandwidth more than 28.9 Mbps 2020OpenSignal . Further, as websites become more complex they may act differently in networks of varying performance. Considering these differences, does the adage of network latency being the limiting factor website performance continue to hold? To answer this question, we take a fresh look at characterizing mobile website network performance requirements. We propose a new website characterization method we dub the Critical Path of Improvement (CPI), illustrated in Figure 1 . CPI allows us to understand how small changes to network bandwidth and latency affect website performance as measured by the Perceptual Speed Index (PSI) Gao2017Perceived , or any other web performance metric Furtak2019ODDness . We base CPI on the observation that at any given point of network performance on the bandwidth/latency plane, PSI will improve on average with small increases to bandwidth, or small reductions to latency. Figure 1 illustrates PSI improvement by lighter points along the CPI. We evaluate the impact on PSI of these small changes in network performance and record CPI as the path of the fastest PSI improvement (details in Section 3 ). Effectively, CPI allows us to understand when a website performance benefits more from additional bandwidth or from lower latency. Figure 1 : Illustration of CPI and network conditions. To understand whether latency or bandwidth of mobile network is the limiting factor for a given website, we analyze the relationship between a network’s performance and the CPI. In Figure 1 , the rectangular regions represent network performance envelopes in terms of minimum and maximum latency and bandwidth measurements. For a network whose performance envelope lies in region D, the path to better PSI points downwards to the CPI and so via lower latency. On the other hand, if the network performance envelope corresponds to region C, the path to the CPI and better PSI lies towards higher bandwidth. We want to understand how common are scenarios where a website could benefit from higher bandwidth, or lower latency, than what is available in a given network deployment. In other words, how often is website performance limited by insufficient bandwidth versus by high latency. This paper offers the following contributions: • We propose CPI – a new website performance characterization method. We show how developers may use CPI to understand the impact of small changes in network performance on the performance of their websites. • We analyse the CPI of 45 websites from the Alexa-100 list in four mobile networks across 57 US cities to identify cases where websites are bandwidth- and latency-limited. • We analyse regional variation in network performance to show areas of the US where network bandwidth tends to be the limiting factor to website performance. • We point to factors of webpage design that contribute to websites being limited by bandwidth. • We offer a discussion of how developers and network operators may use these results to improve the match between website design and network performance. The rest of this paper is organized as follows. In Section 2 we discuss related work on website performance bottlenecks. Section 3 details CPI measurements and comparisons with network performance. In Section 4 we describe our measurement methodology and discuss measurement results. Finally, we offer a discussion of the applicability of our findings to website developers and network administrators in Section 5 and conclude in Section 6 . 2 Related Work A number of papers quantified performance bottlenecks in mobile websites. Nejati and Balasubramanian found that device processing speed as well as delays in transferring objects on the critical path of rendering are the main causes of lower website performance on mobile devices Nejati2016Depth . While some researchers have focused on reducing the computational load of rendering Pourghassemi2019What ; Ren2018Proteus , others have focused on eliminating loading bottlenecks. One approach to eliminating loading bottlenecks is to reorder objects on the critical path Liu2017SWAROVsky ; Furtak2019ODDness . Even with object reordering, Furtaket al. have shown that increased bandwidth improves page Speed Index by permitting large images to load more quickly Furtak2019ODDness . However, others have observed that the impact of extra bandwidth is less pronounced on Page Load Time (PLT) Rajiullah2019Web . Hanet al. have shown that extra bandwidth improves the performance of large pages Han2016When . They loaded pages using MultiPath TCP (MPTCP) over cellular and WiFi connections in parallel and showed that the loading process can use the bandwidth of the additional link effectively, when the performance of the links is comparable in terms of latency. These results, however, focus on the impact of network performance in general, but do not consider the correspondence of specific web pages

Read the complete text