Featured Interview:
Breaking the Speed Limit: Google’s Performance Guru on Accelerating Web Performance
Steve Souders is the author of High Performance Web Sites, Essential Knowledge for front-end Engineers, and creator of the YSlow front-end performance tool. He spent a number of years at Yahoo!, where he was the Chief Performance Yahoo! Currently, Steve is a member of the technical staff at Google, and a frequent speaker on Web performance topics. Benchmark talked to Steve to get his perspective and tips on optimizing front-end Web site performance.
Benchmark: You’ve been on both sides of the performance equation—the back-end and the front-end—and you have an expert point of view on how each impacts the ultimate user experience. Where do you get the best performance gain for your effort?
Steve Souders: I break performance into two main areas: efficiency and response time. On the efficiency side, we’re looking to build Web application architectures that can scale, that can handle a huge number of users and requests and large amounts of data in an efficient way, while also bringing in our hardware costs and data centers, space costs, power costs, of course maintenance on all that hardware. This is all obviously on the back-end.
Until I got more focused on it, the thought was that this was also the place for improving the response time—making the user experience faster. Prior to starting my work in the performance area about four years ago, I was running large Web applications and people complained about response times. So I would look what could be done to the back-end architecture to try to bring up a 200 millisecond page creation time to 150 milliseconds or something like that.
But it turns out that, if you view a Web page loading from the end user’s perspective, we’re talking about thousands of milliseconds, and so this back-end timesaving of 100 or 200 or 500 milliseconds really isn’t much of a factor when you look at the overall load time of a page from the browser’s perspective, from the end user’s perspective.
READ FULL ARTICLE |
