More Bandwidth Doesn't Matter (much)

2020-11-01

Mike Belshe

2010-04-08

“given the way that HTTP uses short, bursty connections, it turns out that round-trip times dominate performance more than bandwidth does”
“web pages are comprised of many short connections”

Two Factors in Web-page Downloads: Bandwidth and Round-Trip Time

“If we make an analogy between plumbing and the Internet, we can consider the bandwidth of the Internet to be like the diameter of the water pipe. A larger pipe carries a larger volume of water, and hence you can deliver more water between two points.”
“no matter how big your pipe is, if the pipe is empty, and you want to get some water from one point to the other, it takes time for water to ravel through the pipe. In Internet parlance, the time it takes for water to travel from one end of the pipe to the other and back again is called the round trip time, or RTT.”

A test in which page load time is measured for bandwidths in {1,2,…10}Mbps for a single page not different between the tests, with results visualized in multiple charts.
“As you can see, there are diminishing returns as the bandwidth gets higher”
“According to the Akamai Internet Report Q2’09, the average bandwidth in the US is only 3.9Mbps.”
From 1-2Mbps, and 2-3Mbps are 35 and 15 percent decreases in page load time, but “[a]n increase from 5Mbps to 10Mbps amounts to a 5% improvement in Page Load Times”
The last of the charts “shows the effective bandwidth of the web page download as the raw bandwidth is increased”.
“At 1Mbps, web pages can be downloaded at ~69% of the total bandwidth capacity. At 10Mbps, however, web pages can only be downloaded at ~16% of the total capacity”
In other words, if you pay for extra-wide pipe, most of that extra width is unused.

“For this test, we fix the bandwidth at 5Mbps, and vary the RTT from 0ms to 240ms. Keep in mind that the worldwide average RTT to Google is over 100ms today.”
In the results, Page Load Time is reduced nearly linearly with RTT
“In these tests, with bandwidth fixed at 5Mbps, each 20ms yielded 7-15% reduction in PLT.” -" With high RTT, [effective] bandwidth of a page load was as low as 550Kbps, which is a little more than 10% of the 5Mbps theoretical throughput. With low RTT, web page downloads still only achieve ~54% of the total bandwidth available. This is due to other factors of how web pages are loaded."

“If users double their bandwidth without reducing their RTT significantly, the effect on web browsing will be a minimal improvement. However, decreasing RTT, regardless of current bandwidth always helps make web browsing faster. To speed up the Internet at large, we should look for more ways to bring down RTT. What if we could reduce cross-atlantic RTTs from 150ms to 100ms? This would have a larger effect on the speed of the internet than increasing a user’s bandwidth from 3.9Mbps to 10Mbps or even 1Gbps.”
“Another approach to reducing page load times would be to reduce the number of round trips required per page load. Today, web pages require a certain amount of back and forth between the client and server. The number of round trips is largely due to the handshakes to start communicating between client and server (e.g. DNS, TCP, HTTP), and also round trips induced by the communication protocols (e.g. TCP slow start). if we can improve protocols to transfer this data with fewer round trips, we should also be able to improve page load times. This is one of the goals of SPDY.”
“it should be noted that this study focuses solely on fresh connections to a site. Continued browsing through a site, so long as it can reuse existing connections, may experience better utilization of the link.”