Built for Speed

Given the importance of load times on revenues, we might ask what causes them to be longer than desirable. There are dozens of rules of thumb, such as “Don’t dynamically scale images in the browser.”

Generally speaking, the end-to-end time is based on time spent:

1. On the client endpoint—smartphone, tablet, laptop, or PC and its browser or app
2. In the network, which is a function of network latency (i.e., delay) and the number of round trips across the network
3. In the cloud or at the corporate data center

To reduce the first is usually a matter of user interface design (and increasing the computing power of the endpoint device, if possible); to reduce the second is a function of reducing the distance between the client and the cloud and reducing the number of sequential round trips; and to reduce the third is a matter of improved application design and improved architecture, including the use of parallel processing.

To better understand the issues surrounding the number of round trips, it’s important to understand how Web pages load. One key factor is the total size of the page, but of perhaps greater relevance is the number of objects, which tend to be images, and thus the number of image requests.28

According to Web performance optimization company StrangeLoop Networks, the average page size is now over a half a megabyte and contains about 80 individual objects, such as banner ads and illustrative graphics and logos.29 The more objects that need to ...

Get Cloudonomics: The Business Value of Cloud Computing, + Website now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.