Google has made Page Speed available online, enhancing it for analyzing web pages targeted at smartphones.
Page Speed was first released as a Firefox add-on running inside Firebug. A week ago it was added to Chrome, being available in Developer Tools if Experimental Extension APIs are enabled in the “about:flags” page. Page Speed for Chrome results have been localized in 40 languages and it is used by 1.4M unique users, according to Google.
Google has just taken a step further by making Page Speed analysis available online, including support for mobile pages. Anyone can test the performance of a web page by providing its URL, if the page is accessible from the Internet without a password. A server at Google will upload that page and verify it against 31 rules grouped in a set of 6 best practices describing what can be done to maximize the performance of a web page:
- Optimizing caching
- Minimizing round-trip times
- Minimizing request overhead
- Minimizing payload size
- Optimizing browser rendering
- Optimizing for mobile
Among others, these practices contain guidance for improving the speed of pages designed for mobile devices, such as deferring the parsing of JavaScript code until it is to be executed and caching landing page redirects. Also, the analysis takes into consideration other factors, such as CPU and mobile network speed. For mobile tests the page is loaded by a WebKit renderer.
Google also provides a set of additional related tools. mod_pagespeed is an Apache module which optimizes the pages delivered and their resources based on a number of filters. For web servers other than Apache, developers can use Page Speed Automatic, a C++ library providing the same functionality as mod_pagespeed. Page Speed SDK enables to integrate Page Speed in other tools.
WebPagetest.org, based on Page Speed SDK, enables developers to test the performance of web pages loaded with different versions of IE in different globally-spread locations in order to have an idea how a page is perceived across the world and where it spends most of its loading time.