First of all thank you @Sven
for adding the proxy IP to the download failed and url resolve failed messages. It makes it possible to compare the performance of proxies. As you said in another thread not all of the error messages are caused by the proxies and some of them may not be, and even if the messages are not a direct sign of performance they are at least a halfway decent indicator of proxy performance. (one of the few we have in fact)
After the gmail dottrick vs hotmail test
I wanted to compare semi-dedicated proxies with dedicated proxies, so I set up a 6 hour log to see how many "download failed" and "url resolve failed" messages I would get and if there was any difference between semi-ded and dedicated proxies.
19 proxies - 14 semi-dedicated, 5 dedicated
120 Mb/s connection
Test Run Started: yyyymmdd,hh:mm - 20140319,00:05
Test Run Finished: yyyymmdd,hh:mm -20140319,05:59
total events in log: 153289
download failed events in log: 10820
url resolve failed events in log: 1037
14 semi-dedicated proxies failed 8252 times
which means an average of 589 fails in a timespan of 6 hours average per proxy (see details here
5 dedicated proxies failed 1410 times in a timepan of 6 hours
which means an average of 282 fails per proxy in a timespan of 6 hours
Conclusion: Semi-dedicated proxies are involved in twice as many error messages per proxy as dedicated proxies. It seems like there is a difference between semi and dedicated proxies after all. Also the number of proxy fail messages with IP does not match the number of fail events - so this is something to look into, but will take more time.
What do you think? Is it worth getting dedicated proxies? Why? why not?