spped checkers

I have recently 'boosted' my connection speed with talktalk to 16mb, and after their staff tweaked something, I was getting speeds approaching that using their speed check. This morning my speed with their speed check is 10782 502 upload, but on your site your speed checker shows 7582 and 772 upload. Which is the more realistic, factual or is it because the tests are running peak performance compared to average or something such as this.

Open task manager and select the Networking tab.

Then run the two speed checkers one after the other and then look at the display. Did one download appear to run at a faster rate that the other? Make a visual approximation of the speed achieved for each of the two blocks of data.

I always rely on the TBB checker - it always appears accurate and consistent whereas others measure a peak or cached performance and over a short time frame which can give an inaccurate result. There are some sites which tell me I am getting 50Mbps - not bad on an up to 8Mbps connection that syncs at around 3.7Mbps

In reply to a post by peter_lord:

I have recently 'boosted' my connection speed with talktalk to 16mb, and after their staff tweaked something, I was getting speeds approaching that using their speed check. This morning my speed with their speed check is 10782 502 upload, but on your site your speed checker shows 7582 and 772 upload. Which is the more realistic, factual or is it because the tests are running peak performance compared to average or something such as this.

What is your actual sync speed? As a general rule of thumb effective throughput (ie;the speed at which data gets from the remote server to your hard disk) will be 85% of sync speed. If you have genuinely got a 16Mb connection (ie;not just 'up to 16Mb') then you should be seeing around 13.6Mb throughput.

TBB's tester is one of the better ones but I sometimes get naff readings from it even though other sites and my own network monitoring indicate no problems. Probably it's just the link between my ISP their site.

Aside from the 85% rule above you should develop a collection of testers and use them fairly frequently. What you're looking for is a straw-poll. The fact that one tester gives a particularly good or bad result is irrelevant. It's the concensus that matters.

You should also consider the time of day. In fact in a lot of ways this is more important. Run your tests around 8am and 8pm. A significant drop in the latter indicates an oversubscribed ISP. A significant drop in the former indicates something odd going on

Finally it's always worth noting that all a tester ever does is tell you the speed between you and the tester. Unless that is the site you actually want to access (highly unlikely) then the results are dubious. Testers are good for getting a general feel for the state of your connection and ISP but what really matters is how fast you can exchange data that matters with servers you care about

Edited by Andrue (Thu 30-Jul-09 11:51:27)