Slow "Updating" in Minkata.71

[Mac_Fife]

Personally, I think the checker that Treehugger used gives results that are on the slow side if you're on a faster connection (Cable, Fibre) and thinkbroadband's checker seems to be more popular for comparisons across UK ISPs. However, I usually use SpeedTest.net and just hit the "Begin Test" button which chooses the best test server based on ping (and has servers around the world).

At this point, I can't really imagine what might have changed to cause Treehugger's startup time to increase by 3-4 times while I can see no real difference . I'm inclined to say it's maybe a case for using Wireshark to try to see what's going on with the network traffic.

[Treehugger]

I just tried all three speed checks and got very different results. The original one gave 15.39; the second one Mac mentioned gave 54.25; and his preferred one gave 29.81. Sigh! Anyway, looks like I have a pretty fast connection now.

[sut]

OK thx will look into all

[Mac_Fife]

Different results from different speedtesters is rather to be expected. They're only an indicator, and you'd need to try to control lots of variables to get consistent results. It's not worth the effort: We've established that raw speed is unlikely to be a factor here. Since I can get Minkata started in roughly the same time as MOULa (and roughly the same as Treehugger's MOULa startup) then it's not a particular limitation of the server either. Both Treehugger and myself are in the UK, so that in itself is not a factor.

My gut feeling is that some host along Treehugger's route to the Minkata server is dropping a high percentage of packets - that's an issue that could suddenly arise because of a fault, persist for a while, then suddenly disappear again if the affected machine gets restarted.

[rarified]

Also with speedtest.net, you can choose the server you want to measure the network bandwidth with.

You might use speedtest.net (Ookla, I believe) and choose a server from either Denver or Colorado Springs, Colorado and see what the throughput to our near vicinity is.

_R

[Mac_Fife]

On the speedtest.net map, I get one 'dot' that lists two servers in the Colorado Springs area plus one in Woodland Park and another 'dot' with four around Denver plus one in Aurora. Of two that are listed as Colorado Springs, Kellin Communications actually seems to be the North side of Monument, so is about mid-way between Colorado Springs and Denver. The other, Data102, seems to be in downtown Colorado Springs (102 South Tejon Street) so is fairly close to rarified's Foundry.

Of course, physical proximity is no guarantee of comparable performance since other factors will determine the overall "route" that data takes, but it can give an indication of what to expect.

[Treehugger]

tried the test with data102 and got download speed of 22.86. Tried again letting it choose for itself and got 27.17. I've posted up the trace with my ticket.

[sut]

Boy wish I had those download speeds but dsl is not that fast ! my check was 2.6 mb/s , and when I timed Minkata I got about same 53 sec but when i tested Moul I got 65 sec Humm ! just thought you might like to see this

[Mac_Fife]

Thanks sut. It's an interesting comparison because it demonstrates that a much slower download speed doesn't have a huge influence on how long it takes to start either Minkata or MOULa.

[Treehugger]

Boy wish I had those download speeds but dsl is not that fast ! my check was 2.6 mb/s , and when I timed Minkata I got about same 53 sec but when i tested Moul I got 65 sec Humm ! just thought you might like to see this

thanks for testing, sut - clearly, speed isn't everything!