Hello, What you measured is only the latency of the network, this has a relation to the bandwidth usage, but it is not the main factor.
To really know how fast each link will be, your only recourse is to actually try to download and see the speed. You need to give TCP some time to stabilise its network congestion protocol to know what is the cruise speed. The time for this depends on many factors, among which latency is an issue (RTT, Round Trip Time, is the issue actually). To understand why, consider a very close mirror on the IIX, but it's connection to the IIX is a dialup 9600bps modem, and another mirror is in the US with a OC-whatever connection as the smallest link between you and the US mirror. The local mirror has a very short latency (for the nit-pickers, it's a VERY fast modem, that waits most of its time), the remote mirror has a very long latency. However when you will try to actually transfer something the local mirror is bounded by it's low bandwidth more than you gain by the low latency. Baruch * Amir Hardon <[EMAIL PROTECTED]> [040525 16:56]: > I tried to find out which Mandrake mirror will be the fastest for me before > doing a network install, so I downloaded the mirrors file(well actually only > the relevant part of it for me) and hacked up this line: > > sed -n 's/^.*\(http:\/\/\|ftp:\/\/\)\([^\/]*\)\([^ ]*\).*$/\2/p' mirrors.html > | fping -e | sed -n 's/\([^(]*\)(\([0-9]*\).*/\2 \1/p' | sort -n > > I started downloading the floppy image from the mirror that replied to my ping > the fastest, but the download speed was very low... > > I tried to send bigger pings (of 1000 bytes) and still found no relation > between the ping reply time and the actuall download speed... ================================================================= To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
