Speed limit?

Hi,

I'm on host22. I recently have realized that both POP3 and SFTP (I have no FTP but SFTP) don't go over 14Kb from my server to my home computer.

I don't know if there is any parameter in my Gentoo that can limit this speed, but I don't remember touching anything.

Is there a known issue of Linode? Is it right? My home connection is a 4Mb ADSL and I get more than 300 Kb of downloading from sourceforge so I think the problem is not in my side.

I have plenty of transfer this month: "You have used 5% of your monthly transfer"

Host is not overloaded: "Host Load: medium"

Also, IO tokens are ok:

iocount=11403254 iorate=132 iotokens=399839 tokenrefill=512 token_max=400000

Load seems to be ok too:

load average: 0.32, 0.46, 0.42

(All these parameters taken when downloading was in progress)

Any ideas? Should I open a support ticket?

Thank you in advance!

3 Replies

There may be many issues here. I am in Sydney, Australia, and between my workstation here and my linode virtual server in the states I can get over 1MB/s transfer on host24 (using rsync to mirror disk images).

A good tool is iperf (http://dast.nlanr.net/Projects/Iperf/). Using this you can ask your linode server (running iperf) to blast UDP packets out at a given rate (say 100KB/s initially) over a certain time period (say 30 seconds) and running iperf on your home computer (it can be Win32) you can measure how many of those UDP packets sent make it to your home computer.

Often UDP can be transferred at a greater rate than TCP. Why? Because TCP connections are rate limited by the protocol stack in either end of the connection. It is possible that due to distance or delay or many other factors between you and linode TCP is thinking it should run slower.

Have a good look around the iperf website above, it may help explain some of the potential performance issues you might be having.

I'm also in Sydney, Australia.

I get as much as my cable can handle, about 300KB/s on a completely random file through HTTP.

I'm on host 2 though. I think the problem might actually be on your side. Have you tried offering a file with something as simple as netcat?

Sometimes what happens is that you find a bottleneck where the route crosses over between backbone carriers. For instance, from my job, if I SSH to my my systems behind a cable modem at home, and then from there to my linode, it works great. If I SSH to our company's datacenter, and then to the linode, it works pretty well, but not quite as well (slightly more latency). If I SSH straight from the office without an intermediate hop, the latency spikes are abysmal most of the day. Looking at the traceroutes, the jump seems to happen where it crosses over to another carrier.

YMMV.

Reply

Please enter an answer
Tips:

You can mention users to notify them: @username

You can use Markdown to format your question. For more examples see the Markdown Cheatsheet.

> I’m a blockquote.

I’m a blockquote.

[I'm a link] (https://www.google.com)

I'm a link

**I am bold** I am bold

*I am italicized* I am italicized

Community Code of Conduct