Quote:
Originally Posted by Steve1888
The power levels were fine, they didn't change in any significant way during the low speed indications and the high speed.
Granted about the accuracy of the speedtests, and if these had been random then I would have simply dismissed the results. What perked my interest was timing of the less than random changes in the test results.
You can't be too paranoid... well maybe.
|
Just climbed into this thread without reading all the responses. But I hope what I've got to say is helpful.
You obviously know what you're doing, especially spreading your tests around different servers. As I've said in other threads,
Speedtest.net is unreliable IMHO.
I did what you did and in a c. 1 minute period did three tests at Maidenhead, London & Rome followed by two more tests immediately at Rome. (I'm on a 20 Mbps servive using a Homeplug bridge which keeps me down to c. 15 Mbps).
Maidenhead: 3.2 Mbps
London: 9.8 Mbps
Rome: (1) 13.2 (2) 10.9 (3) 7.9 Mbps
This told me what webcrawler250 has already told you. Namely (but in other words) that on the internet, anything could be happening at any time anywhere in terms of linbe or server contention.
I noticed also that my ping time to Rome was the same as to London (same number of hops).
This is a common pattern. One of the constants in this is the uselessness of the Maidenhead server. It always depresses results and that is why you were completely right in spreading your tests.
The wise men here correctly say that downloading a very large file or even two concurrently will provide a better picture over a longer snapshot period. These longer periods will more typically represent the instant changes that occur across the internet and the VM system.
It is clear to me from what you'gve written that you are "victim" of this instantaneous change when you call VM - a few seconds later, and the picture could have been entirely different.
So Speedtest.net and its ilk might be a tempting starting point but the acid test and basis of decision for clling VM when you think things are going slowly is to measure the time it takes to download a large file (say > 250 MB) and then to do it at least twice at different periods of the say. Sort of thing.