Very slow downloads from highland3.seas.upenn.edu server

[Schrödinger's cat]

My last two work units from highland3.seas.upenn.edu (158.130.118.25) were both 92MiB, and took six minutes each to download. An 80MiB upload occurring simultaneously with the first download took 12 minutes! A non-simultaneous 80MiB upload took just two minutes and 15 seconds.

I believe the slow downloads and uploads are due to the server, as a 30MiB download from fah1.innovatr.ca took just 13 seconds, while a 31MiB upload to the same server took 50 seconds.

[muziqaz]

My last two work units from highland3.seas.upenn.edu (158.130.118.25) were 92MiB, and took six minutes to download. A simultaneous 80MiB upload took 12 minutes!

I believe the slow downloads and uploads are due to the server, as a 30MiB download from fah1.innovatr.ca took just 13 seconds, while an upload to the same server to 50 seconds.

You are in Australia, you have to account for the traffic between you and East coast USA, which is nearly across the world

[Schrödinger's cat]

You are in Australia, you have to account for the traffic between you and East coast USA, which is nearly across the world

That doesn't explain why one server is very slow compared to another. Also, data travels very quickly over fibre optic cable, even across oceans.

A Speed test to from Victoria, Australia to Portland, Oregon, U.S.A. gives a download of 27.33 megabits per second, with a ping time of 183 milliseconds and a download latency of 213 milliseconds (internet in Australia is expensive for singles!).

[muziqaz]

You are in Australia, you have to account for the traffic between you and East coast USA, which is nearly across the world

That doesn't explain why one server is very slow compared to another.

A Speed test to from Victoria, Australia to Portland, U.S.A. gives a download of 27.33 megabits per second, with a ping time of 183 milliseconds and a download latency of 213 milliseconds (internet in Australia is expensive for singles!).

It does. You do not have direct line to upenn server Even connection to the same server can vary every minute.

[Schrödinger's cat]

Sixteen of the last eighteen 92MiB downloads since April 17, from highland3.seas.upenn.edu, took between 28 seconds and 54 seconds, with an average time of 36 seconds, except for the last two downloads.

So, the download time has increased by 6.6 to 13.3 times!

All downloads were 18260 work units.

Finally, a 92MiB download from Portland, Oregon, U.S.A., should take just 28.25 seconds on my internet connection (@ 27.33Mb/s).

[gordonbb]

I'm now seeing slow downloads and uploads on highland1, 2 & 3 across 5 Slots in 3 systems. Sometimes taking an hour to download a WU.

Code: Select all

traceroute to highland1.seas.upenn.edu (158.130.118.23), 30 hops max, 60 byte packets
 1  * * *
 2  lo0-0-bng02-tor1.teksavvy.com (206.248.155.247)  13.678 ms  14.346 ms  14.335 ms
 3  ae7-0-cr01-tor1.teksavvy.com (206.248.153.36)  16.241 ms  16.228 ms  16.499 ms
 4  ae2-0-bdr01-tor1.teksavvy.com (206.248.153.9)  14.999 ms  14.987 ms ae1-0-bdr01-tor1.teksavvy.com (206.248.153.3)  14.958 ms
 5  toro-b5-link.ip.twelve99.net (62.115.171.248)  15.807 ms  16.134 ms  16.121 ms
 6  lumen-ic-361864.ip.twelve99-cust.net (213.248.94.123)  16.389 ms  12.854 ms  12.794 ms
 7  * * *
 8  UNIVERSITY.edge5.Newark1.Level3.net (4.16.91.150)  26.747 ms  28.839 ms  27.238 ms
 9  hnt-brdr-dmz.hnt-fw.router.upenn.edu (128.91.48.6)  28.490 ms  28.473 ms  28.457 ms
10  hnt-fw.hnt-brdr.router.upenn.edu (128.91.48.14)  28.226 ms  28.202 ms  26.940 ms
11  hnt-brdr.vagelos.router.upenn.edu (128.91.244.34)  26.925 ms  26.857 ms  27.131 ms
12  core-router.seas.upenn.edu (158.130.0.249)  55.197 ms  55.862 ms  54.544 ms

Note the almost doubling of latency (RTT) on the last hop of a traceroute

Latency to all three systems seems within normal bounds:

Code: Select all

--- highland1.seas.upenn.edu ping statistics ---
13 packets transmitted, 13 received, 0% packet loss, time 12023ms
rtt min/avg/max/mdev = 27.582/30.135/38.385/3.518 ms
--- highland2.seas.upenn.edu ping statistics ---
10 packets transmitted, 10 received, 0% packet loss, time 9016ms
rtt min/avg/max/mdev = 27.489/28.355/31.920/1.252 ms
--- highland3.seas.upenn.edu ping statistics ---
26 packets transmitted, 11 received, 57.6923% packet loss, time 25360ms
rtt min/avg/max/mdev = 27.521/30.420/37.710/3.514 ms

[toTOW]

Highland servers have always had random download/upload issues when you're not in the US and no one never figured out why ...

Hihgland3 has been updated to the latest version of the WS software, it might trigger new issues.