Before this, I'm using shared 100Mbps bandwidth. Its fast enough.
And now, the server running dedicated 10Mbps bandwidth.
When running 10Mbps, it takes more time to completely load the same page.
The server bandwidth usage is small, with average less than 5Mbps.
(I can see some website hosted at same data center loads very fast.)
It is probably because the new line has higher latency. I recommend you go read "It's the Latency, Stupid" (Not calling you stupid, that happens to be the title.)
Lots of webpage requests are going to be small, so latency becomes a significant factor because it is the initial delay in the request and delivery of a page.
How are you deciding how quickly the transfer is occurring?
Note that the transfer speeds that you notice are limited by the slowest link in the chain between you and the server. If you are testing transfer rates by loading a page in your browser, you're probably being limited by whatever network speed you have available through cable/DSL/whatever.
If you notice that another website hosted at a server at the same data center is loading faster, then perhaps bandwidth isn't the trouble. Your browser could be rendering your page more slowly than it renders the other page.
For investigating the performance of your page, take a look at http://developer.yahoo.com/yslow/, there are other similar tools available.
Deciding on speed by when a page is rendering isn't a good technique. The content of the page affects the rendering time. There are a variety of ways to actually time the transfer, but again, that's probably telling you more about your Internet connection than your servers.
This might be completely unrelated to your bandwidth cut-down.
Have you got any network diagrams? Probably you got more requests since then or your webservers keep-alive settings are set too high, limiting new connections and some bandwidth.
Depends on a lot of factory.
For a large page, yes, it will show slower. X bytes take longer going through a 10mbit line than through a 100mbit line.
That you use an average of 5mbps is irrelevant - you do not talk of averages here. The question is not so much the average, as the spikes, and the spikes can not be handled that fast on a 10mbit link than they can on a (reasonably fee) 100mbit link.
