End of November 2009 Google announced the download speed is getting more important. Google's ranking algorithms are going to include the download speed as one factor for ranking websites. You can check Download speed for your own website, pictures are not included in this test.

Because people are getting more and more "frenzied" when surfing on the net the surfers want everything immediately. We can therefore soon expect to see slow downloading webpages to slide down wards in the search engine results pages ( SERPs ) being replaced with faster downloading web pages.

Connection Speed

The speed is counted in bytes/sec. or KB/sec. (1 KB = 1024 bytes) and is specific for each computer and connection. Depending on the traffic volume at different Internet Service Providers ( ISP ) the speed is variable. It is higher or lower at certain times of the day/week. You can check the connection speed for any computer / connection at browserhawk.com/. With a broadband / DSL cable connection you can expect about 20 - 30 KB/sec. or even considerably more, while a modem + telephone line connection often is as low as about 2 KB/sec.

The connection speed you get is a theoretical one and is measured by sending a certain quantity of data over the connection you are using. Which way and via which intermediate servers / ISP's your connection goes each time your browser connects to an internet website is out of your control. The greater number of intermediaries the lower is the connection speed. You can use the connection time value to rank different computers / connections in your hometown.

To get the real total display time for a certain page you must, however, use a stopwatch.
The real total display time can be divided thus:

1. The time needed for the browser to find the server requested.
   
2. Time needed for the server to answer and your browser to ask for a certain page. When you want the browser to ask for "index" page first, add a slash (/) after domain name. The slash is short for "index.html" in Computerese [ or .htm or default.html or whatever the entry page is called ].
   
3. Actual time needed to download the code, including pictures. Note that each picture / external file requires a separate download request by the browser.
   
4. Time required for the browser to process the code and to display the result.

See below how to measure total display time.

Download Time

There can easily be much more than 100 customers at the same server at any one time, each one downloading from a different site.  To optimize the server's service, each page code is split up into packages delivered as a train with several wagons. One can imagine the process as follows.

If the file requested is large, the file itself can easily be split into several packages, as, for instance, "interlaced .gif" pictures, which are delivered by and by in several slices, very often 3 slices.

Each page code is sent in packages so that each customer gets a small package at his turn. Between each package coming to your computer there can then be a hundred or more customers receiving their small packages. That's why the download time gets long when the server is busy (= very many customers in line). The server is optimizing the total service time for all customers at each point in time.

Diagram 1.

The arrows (in Diagram 1) starts at a point in time, and shows the flow of time as one imagines it. The request in the middle can be imagined as being sent to your computer.

The browser has to request each file, like external files (.css or .js), separately, same as with picture files. If the external file is your style sheet may be it comes in two packages and your text cannot be displayed before these files have been processed.

As the text is displayed first directly from the downloaded "main" code and the external files may be come later, it will have an effect of parts of the page like floating around and looking for their final resting place. While this goes on your visitor has to wait because he/she cannot start reading before the process is done.

The modern browsers do save the style sheet files in the cache, although Explorer 6 sometimes refuses to up-date a saved style sheet. So how to know what code to use and when? When choosing between CSS and HTML, ask yourself if you can get same or almost same display with simple HTML. If you can, then use HTML because the code is shorter and therefore the download time is also shorter.

It also explains why it is important to indicate to the browser the physical size of the pictures ( in pixels ). When that is specified your browser simply reserves a place for each picture and displays the text around it while still downloading the picture code.

When you arrange some text in the beginning of the your page as well as between the pictures, your visitor have something to occupy himself / herself with (= reading the text ) while the picture(s) are downloading.

Display Time

Once your browser has received a package, it starts processing the data and at the same time downloads the next package etc. To a human it seems like the browser is doing 2-3 different things at the same time, which is in fact exactly what the browser is doing.

Diagram 2.

The process of receiving your files could look like in Diagram 2 above.  "Text #" and "Pic #" are the separate requests received from among the middle of other customers. Text #1 can be imagined as being your request from Diagram 1 above.

Diagram 2 explains why external files (.css or .js) increases download time. The browser has to request each file separately, same as with picture files.  If the external files are your style sheets may be they come in two packages and your text cannot be displayed before these files have been processed.