How important is it to write on a site with high traffic?

Question

Some high traffic sites seem to have almost crazy images with images with almost one large image.

How big is the difference with high traffic and the complexity of service in the long run?

I know this may be a question of “how long has it been part of the line,” but I’m interested in understanding any performance improvement experience people have gained from such an exercise.

Answer 1

A few years ago, I made a few digits for a minimum .GIF size of 43 bytes. To send this 1x1 image, 246 bytes of the HTTP protocol were sent with it, not to mention that at least for the original request (moreover, if there is also a cookie). There is also TCP / IP overhead. Besides all this, this is a round-trip time. The speed of light is still finite, and it must be taken into account, especially for the overhead of several queries.

In short, fewer loaded items equals faster site. Even if the final image of the sprite is larger than all the images in combination, you still win mainly by reducing the number of overhead sets and round trips.

Answer 2

Using sprites, you reduce the number or HTTP requests from the server.

Images make up the vast majority of HTTP requests, and Yahoo estimates that about 80% of the time spent loading page content is spent on unnecessary HTTP requests.

If you have done something as simple as the homepage of amazon.com, there could potentially be more than 50 images that will be uploaded every time a user (without a saved cache) gets to his homepage.

If you think that Amazon should have millions of users accessing their site every day.

Having created a sprite from all the images, the user needs to upload only one file instead of fifty.

Reduce server response time and improve end user experience.

Answer 3

I believe that most of the overhead is just an image request. Obviously, loading images takes more time, but initiating a request, waiting for a response, and downloading each image can add up a significant amount of traffic over time. The same goes for compression.