Google SPDY Results
Exactly four years ago, Google announced the protocol SPDY, which was conceived as an upgrade for HTTP 1.1 in order to significantly increase the speed of all types of connections. SPDY allows you to halve latency when working through HTTP. This is done through three methods: 1) multiplexing requests; 2) prioritization of requests; 3) HTTP header compression.
The first "laboratory" tests of SPDY showed an increase in the speed of loading web pages by 55% , in mobile networks - by 23% . However, independent tests on real sites showed no performance gains at all. One reason is that for real sites, resources are loaded from different domains, including those where there is no SPDY support.
Over the past four years, much has changed. SPDY itself is optimized and grown to version 3.1, and it was decided to make it the basis for the next generation protocol HTTP 2.0 . The current implementation is supported in all modern browsers, including Chrome, Opera, Firefox and even Internet Explorer, in dozens of server platforms and on many large sites.
Yesterday, Google published the results of testing SPDY on their own sites. These are not laboratory tests, but real statistics from millions of user sessions with different types of network connections, that is, with different access speeds.
The table shows the change in the latency indicator - the time between sending a request and receiving the first incoming event in the browser. The median value is indicated, the average for 5% of the fastest compounds and for 95% of the rest.
The tests used Chrome 29 and compared the work on HTTPS and SPDY.
Naturally, on other sites the use of SPDY may not give such an effect. The specific result depends on the amount of downloaded content from extraneous resources and on dozens of other factors.
At the same time, Google says it continues to optimize compression, flow control and traffic prioritization algorithms in SPDY. There are a lot of interesting ideas, so by the time the HTTP / 2 standard is adopted, this protocol will work even more efficiently.
The first "laboratory" tests of SPDY showed an increase in the speed of loading web pages by 55% , in mobile networks - by 23% . However, independent tests on real sites showed no performance gains at all. One reason is that for real sites, resources are loaded from different domains, including those where there is no SPDY support.
Over the past four years, much has changed. SPDY itself is optimized and grown to version 3.1, and it was decided to make it the basis for the next generation protocol HTTP 2.0 . The current implementation is supported in all modern browsers, including Chrome, Opera, Firefox and even Internet Explorer, in dozens of server platforms and on many large sites.
Yesterday, Google published the results of testing SPDY on their own sites. These are not laboratory tests, but real statistics from millions of user sessions with different types of network connections, that is, with different access speeds.
The table shows the change in the latency indicator - the time between sending a request and receiving the first incoming event in the browser. The median value is indicated, the average for 5% of the fastest compounds and for 95% of the rest.
Google news | Google sites | Google drive | Google maps | |
Median Average | -43% | -27% | -23% | -24% |
5% fastest | -32% | -thirty% | -fifteen% | -20% |
95% of the rest | -44% | -33% | -36% | -28% |
The tests used Chrome 29 and compared the work on HTTPS and SPDY.
Naturally, on other sites the use of SPDY may not give such an effect. The specific result depends on the amount of downloaded content from extraneous resources and on dozens of other factors.
At the same time, Google says it continues to optimize compression, flow control and traffic prioritization algorithms in SPDY. There are a lot of interesting ideas, so by the time the HTTP / 2 standard is adopted, this protocol will work even more efficiently.