Even on the same machine the throughput between servers varies a lot with out any visible reason
I had a short look at
B060482C784788B8A564DECD904E14CB305C8B38 (your 'slowest') vs DC41244B158D1420C98C66F7B5E569C09DCE98FE ('fastest') running on the same IP.
The relay graphs look like B060 was down for a long period of time between October until a week ago. DC41 was not.
sample lookup https://exonerator.torproject.org/?ip=176.10.104.241%C3%97tamp=2015-10-10
Moritz hinted it might have to do with the way the tor "bandwidth scanners" measure the ability of a server to handle traffic.
Can you explain me / point me to documentation where this process is described
bw scanner spec can be found here:
https://gitweb.torproject.org/torflow.git/tree/NetworkScanners/BwAuthority/R...
on how their vote is processes for the consensus:
https://gitweb.torproject.org/torspec.git/tree/dir-spec.txt#n1890 https://gitweb.torproject.org/torspec.git/tree/dir-spec.txt#n2149
You can download votes from collector to find out how bwauths measured your relays:
https://collector.torproject.org/archive/relay-descriptors/votes/