What tool or method can I use to determine why the download is so much slower at the office when the networks read the same speed.
The network read the same speed towards the metering server.
It might be the case that:
- the metering server is very well connected to both your home and your office
- the company rack is well connected to the ISP serving your home
- but it is not so well connected to the ISP serving your office.
You would need to find another server as close as possible, routing-wise, to your company rack, and verify the actual speed there.
The Internet is made up of lots of interconnected segments, and the speed you will get on any one route will of course be the speed of the slowest segment:
HOME ==== Comcast ======== Company Rack === high speed line
|| . ... slow speed line
fast.com ==++.....................
||
XO ======= Office
Another possibility is that the office firewall is actively limiting the bandwidth towards the company rack, or SSH protocol connections, for some reason.
For example the SSH protocol could be optimized for interactive shell, which is exactly the opposite of what would be required by fast file transfers.
You can check latency and MTU matching between two computers using e.g. winping for Windows or the ping
utility for Unix/Linux.
You can also use the tracert
/ traceroute
utility to display all the hops between home and rack, and between office and rack, and then test the speed of each hop using a ping utility.
Does the office use a domain Server? – Moab – 2016-08-15T22:06:49.977
Regarding the first paragraph 8 MB per second is about 80 Mb per second. Might be easier if all user the same format. – Hennes – 2016-08-15T22:07:27.833
@Moab yes the office is using a domain server – Ethan22 – 2016-08-15T22:12:21.713
@Hennes I was just trying to report exactly what the services/tools showed – Ethan22 – 2016-08-15T22:13:49.747
1Which is a good thing. But I guess I am not the only one doing semi needs less translating in his head. As to possible answer: Any proxy at work (possible with rate limiting - do all other sites cap at the same speed?) Does it speed up if you download more (practical experience at some place: 10MB/sec download when doing one download. If I started 5 I got 5x 10MB/sec downloads. Needless to say I had a lot of concurrent sessions that day :) ).
Also, which scp. SCP isself can be limited depening on random input. (which also shows up per session) – Hennes – 2016-08-15T22:17:32.020
No proxy being used and I can download the same 300 MB file from three different terminals and all are running at 600 KB/s right now. I'll try to set up some tests for downloading from other servers now. – Ethan22 – 2016-08-15T22:22:11.720
Huh, downloading a file from my atlantic.net server shows between 2.5 and 5 MB/s. I guess sometimes you don't think to test something because it doesn't seem like that would be the issue. What kind of issue with Rackspace could exist that would cause issues from the office but not from home? – Ethan22 – 2016-08-15T22:29:23.000
@Ethan22 then ask your domain administrator, we cannot help with domain controlled environments. – Moab – 2016-08-15T22:53:58.367
Technically I am a domain administrator but I hardly ever work on the server. I have been working with the woman that does most of the work and she is stumped right now too. Any direction you can point us in would me nice. – Ethan22 – 2016-08-16T16:37:52.337
Have you tried using a download manager? Sometimes, networks limit the speed of a single TCP connection. If you use something that uses multiple, you may get better speeds. – PulseJet – 2016-08-19T12:22:40.863
What is the latency (RTT) from home, and what from work? scp would be more sensitive to it than plain HTTP. See links at http://serverfault.com/questions/207164/what-does-limit-performance-of-scp as to why
– Matija Nalis – 2016-08-19T19:08:37.150Or perhaps it is simply the case that the destination server is more loaded in different times of day (if your tests from home were carried out at different times of day then your tests from work)? – Matija Nalis – 2016-08-19T19:17:17.770
@MatijaNalis I've tested from home at around the same times as from work so that's not the issue. How could I test the latency? – Ethan22 – 2016-08-20T19:17:57.670
@Ethan22 if your speedtest does not return latency too, you can find manually with
ping destinationhost
. Also, you could rty downloading some file via HTTP ot FTP instead of SCP to compare speeds. – Matija Nalis – 2016-08-20T19:41:42.113@Hennes 8MB/s translates to 64Mbps, not 80Mbps. – var firstName – 2016-08-23T15:37:15.093
If you have no overhead: Yes. With overhead for transmission, padding short frames (e.g. when you press on key in your ssh client and it has to send the minimum frame size of 64) etc etc 1:10 is more realistic. I guess this is a case where: In therory you are right, in practice... – Hennes – 2016-08-23T15:46:28.177
@MatijaNalis RTT from office to server 97.934, from home to server 100.393 – Ethan22 – 2016-08-23T22:46:20.270