Network speed test

Umair

Hello,

I am trying to test speed of a server i got that came with 1Gbps port. There was used to be a script here but I cant find it. Anyone has got a link handy?? Or any other way you can suggest me to test the network speed (upload + download) ??

Server has default install of CentOS.

Thanks

black

wget dl.getipaddr.net/speedtest.sh
chmod u+x speedtest.sh && ./speedtest.sh

Or if you want to use cli for speedtest.net, see https://github.com/sivel/speedtest-cli

tune

Try this benchmark script too.

wget freevps.us/downloads/bench.sh -O - -o /dev/null|bash

Umair

Download speed from CacheFly: 30.6MB/s Download speed from Coloat, Atlanta GA: 16.6MB/s Download speed from Softlayer, Dallas, TX: 38.5MB/s Download speed from Linode, Tokyo, JP: 514KB/s Download speed from i3d.net, Rotterdam, NL: 7.50MB/s Download speed from Leaseweb, Haarlem, NL: 5.77MB/s Download speed from Softlayer, Singapore: 5.81MB/s Download speed from Softlayer, Seattle, WA: 21.2MB/s Download speed from Softlayer, San Jose, CA: 21.2MB/s Download speed from Softlayer, Washington, DC: 58.0MB/s

Doesn't look like 1Gbps by any means

Umair

Ok so I sent a ticket to my provider and they said "Speed test can very because of the network and the other end provider" .... They kinda said speed test from Softlayer doesn't really count.

So, do you have any speed test files who are certain to be connected to 1Gbps port?? Like CacheFly?? Any other files I can use as test ?

Thanks

DalekOfSkaro

@Umair: Is it a dedicated server? If yes, did you try running ethtool eth0 ? Or ethic, depending on what you have.

Umair

@DalekOfSkaro said:
Umair: Is it a dedicated server? If yes, did you try running ethtool eth0 ? Or ethic, depending on what you have.

Yes, it is dedicated server. Doesn't look like an issue with server. (disk performance and Serverbear score was amazing)

Supported ports: [ TP ] Supported link modes: 10baseT/Half 10baseT/Full 100baseT/Half 100baseT/Full 1000baseT/Full Supported pause frame use: Symmetric Supports auto-negotiation: Yes Advertised link modes: 10baseT/Half 10baseT/Full 100baseT/Half 100baseT/Full 1000baseT/Full Advertised pause frame use: Symmetric Advertised auto-negotiation: Yes Speed: 1000Mb/s Duplex: Full Port: Twisted Pair PHYAD: 1 Transceiver: internal Auto-negotiation: on MDI-X: on (auto) Supports Wake-on: pumbg Wake-on: g Link detected: yes

DalekOfSkaro

Umair said: 1000baseT/Full

You have a Gigabit NIC alright. The download speed sure is concerning. Who do you have the server with, and where is it located?

Edit: Let me run a couple of tests myself and I'll share

rm_

Umair said: Doesn't look like 1Gbps by any means

1Gbps port and dedicated guaranteed 1Gbps bandwidth are two different things.
You certainly got the 1Gbps port, and you are able to download at more than 460 Mbit/sec from some locations, and 150-200-300Mbit/sec from others. Are you still not satisfied with that? What's the server price per month that you are paying?

hostnoob

How much did you pay? It's probably 1gbit shared with other servers. Use speedtest-cli posted by black above.

DalekOfSkaro

Testing North America locations
Speedtest from Portland, Oregon, USA [ generously donated by http://bonevm.com ] on a shared 100 Mbps port
    Download Speed: 7.32 MB/sec
    Upload speed: 5.99 MB/sec
Speedtest from Seattle, Washington, USA on a shared 100 Mbps port
    Download Speed: 8.42 MB/sec
    Upload speed: 5.49 MB/sec
Speedtest from Los Angeles, CA, USA [ generously donated by http://maximumvps.net ] on a shared 1 Gbps port
    Download Speed: 12.80 MB/sec
    Upload speed: 4.46 MB/sec
Speedtest from Los Angeles, CA, USA [ generously donated by TeraFire, LLC ] on a shared 1 Gbps port
    Download Speed: 7.37 MB/sec
    Upload speed: 10.63 MB/sec
Speedtest from Denver, CO, USA on a shared 100 Mbps port

Yeah.. That speed test is blah... On a 100 Mbps port using the actual speedtest-cli I get like 100 Mb d/u. Try the one below:

wget -O speedtest-cli https://raw.github.com/sivel/speedtest-cli/master/speedtest_cli.py
chmod +x speedtest-cli

Umair

@DalekOfSkaro said:
You have a Gigabit NIC alright. The download speed sure is concerning. Who do you have the server with, and where is it located?

@rm_ said:
Are you still not satisfied with that? What's the server price per month that you are paying?

This is by no means a "low end server". It came with 1Gbps port so at least should act like one. Does it matter how much I am paying for it? (It should not)

The provider is Hi Velo city

Umair

@DalekOfSkaro said:
Try the one below:

Well Looks like I do have 1Gbps port but may be poor overall network

(They are hosting one server for speedtest so I am getting good speed with it)

But other downloads test (I used HostDim, SoftLayer, CacheFly etc) seems to be kinda poor.

Shoaib_A

@Umair said:
The provider is Hi Velo city

A dedicated 1 Gbps connection "which should act like one" would cost you in 4 figures every month. Bandwidth is not cheap & every provider puts your server on shared ports unless you specifically pay to get a dedicated one.

Shoaib_A

@Umair said:
Well Looks like I do have 1Gbps port but may be poor overall network

(They are hosting one server for speedtest so I am getting good speed with it)

But other downloads test (I used HostDim, SoftLayer, CacheFly etc) seems to be kinda poor.

Same will be the case with every hosting provider's network as speeds are always better when you are downloading/uploading on the same network & location. I see absolutely nothing wrong with network speed of your server.

Umair

@fametel said:

I get that. I am not expecting to always get 100MB/s connection all the time. But at least it got to better than my $5/m vps. However, I am not not getting any stable speed.

From the dedicated server (1Gpbs port )

Download speed from CacheFly: 41.9MB/s Download speed from CacheFly: 30.6MB/s Download speed from CacheFly: 27.6MB/s Download speed from CacheFly: 50.6MB/s

Compare that to the 4 tests I ran on my VPS ($5/m)

Download speed from CacheFly: 71.0MB/s Download speed from CacheFly: 63.9MB/s Download speed from CacheFly: 86.7MB/s Download speed from CacheFly: 91.1MB/s

So do you see my concern ??

Pwner

@Umair said:
So do you see my concern ??

Location also matters.

bsdguy

@Umair

As @rm_ already correctly explained to you "1Gb" can mean different things:
1) a physical 1Gb port on/in the machine (which you have and which, btw, can also be of good or bad quality)

• 2) a connection of your server to a 1Gb switch port
• 3) The full bandwidth of 1Gb at your disposal (and hopefully through a good backend provider).

The first is the standard meaning when a provider sells you a server with "1Gb port". Usually the second comes along with the first. The third is what customers usually understand - and exactly the misunderstanding is often used for marketing purposes.

So, your provider did not lie. You have a 1Gb port (the physical sort). But you - like very many customers - are disappointed because you understood "1Gb" port to have the third meaning.

Usually the volume that comes with a server is a good indicator. If you have, for example, 5Tb in your packet, that would last only for a little more than half a day using a full Gb bandwidth. So evidently every package that comes with less than about 250 TB is not meant for a real full Gb connectivity.

And those limits make sense, because a provider typically pays something in the range of 1.20$ to 2$ per TB volume (small ones usually pay more, larger ones somewhat less and it's, of course also depending on location, but these numbers may help you to get a feeling for the rough dimensions).

Shoaib_A

Are your dedicated server & vps in same location & network? Those 2 things do matter as well.

Edit: @Pwner already beat me to it

Umair

``> @bsdguy said:

As @rm_ already correctly explained to you "1Gb" can mean different things:
So, your provider did not lie. You have a 1Gb port (the physical sort). But you - like very many customers - are disappointed because you understood "1Gb" port to have the third meaning.

You are assuming lots of thing about me. I know the difference and like I said, I was not expecting to have 1Gbps port dedicated for me all the time. However, it should at the very least perform a lil better. (when I compare it with my other servers/vps)
Like in case of Cachefly ... (I tested 3 more links from other providers)

I never said the provider lied or anything like that. I am just merely commenting on the network performance I am getting from them. (Which I did expect to be a lil better than my current provider)

The VPS is from an other provider in Dallas ... The server is in Florida.

Umair

@fametel said:
Are your dedicated server & vps in same location & network? Those 2 things do matter as well.

Edit: Pwner already beat me to it

No, I was just comparing the two results .
I have more results to compare .. the one I gave was a small part of it.

wych

@Umair said:
The VPS is from an other provider in Dallas ... The server is in Florida.

Thats 1,100mi apart... Apples/Oranges.

Ask the provider if they offer a premium or guaranteed connection.

bsdguy

@Umair

Nope. I did assume nothing about you. I merely commented - well noted, friendly - based on what you said and asked. Your own test shows about 500 Mb/s (Washington). This matches with your 1 Gb/s port because evidently a 100 Mb/s port can't do 500 Mb/s.

As you certainly know a whole lot about routing and backend connectivity, too, I see no need for further commenting. regards.

bsdguy

@wych said:
Thats 1,100mi apart... Apples/Oranges.

How strange. So you think that different providers with different backend providers/peering and almost certainly different equipment at quite different locations might actually also lead to different test results?

How on earth could Umair have guessed that?!

rm_

bsdguy said: How strange. So you think that different providers with different backend providers/peering and almost certainly different equipment at different locations might actually also lead to different test results?

His point is that his brand new shiny expensive dedi is overall crappier network-wise than some $5 VPS from another provider, and that's not how it should be (or at least not what was expected).

bsdguy said: Usually the volume that comes with a server is a good indicator. If you have, for example, 5Tb in your packet, that would last only for a little more than half a day using a full Gb bandwidth. So evidently every package that comes with less than about 250 TB is not meant for a real full Gb connectivity.

This is what the OP should look at, if the dedi has "unmetered" or 30TB at 1Gbps, it's no wonder this might be oversold to a higher degree than on a VPS where you most often get only 500GB or 1TB of transfer per month (and hence none of the heavy 24x7 b/w users).

bsdguy

@rm_ said:
His point is that his brand new shiny expensive dedi is overall crappier network-wise than some $5 VPS from another provider, and that's not how it should be.

Oh, I got that. But shouldn't an expert like him know that good as well as bad dedis and VPSs can be at good or less good providers who again can be in good or less good colos and have good or less good connectivity? And, not to be picky, but why should that not be? In my view dedi means just that, dedicated hardware; it say nothing about connectivity (although I see your point that one might be led to assume ...)

Pardon me but I fail to see the connection between VPS/dedi and connectivity.
But then I also fail to see why one would react quite pissed after receiving a not at all unfriendly comment. My fault, no doubt.

Umair

@wych said:
Ask the provider if they offer a premium or guaranteed connection.

Already have. Wondering if it's really worth it. (Did some more tests and it's getting disappointing overall)

Umair

@bsdguy said:
Umair

Nope. I did assume nothing about you. I merely commented - well noted, friendly - based on what you said and asked. Your own test shows about 500 Mb/s (Washington). This matches with your 1 Gb/s port because evidently a 100 Mb/s port can't do 500 Mb/s.

As you certainly know a whole lot about routing and backend connectivity, too, I see no need for further commenting. regards.

Not sure why you getting pissed at. I don't know "a whole lot" or I wouldn't be here asking all that.

The result above was just "part" of the tests I did. I used CacheFly as a reference because generally I have seen better results with it will all providers I have used. And I did multiple tests (just posted random 4 above). While I did get 50+MBps from one location, I would have much preferred better results on average from all locations.

When I am testing "network" I am testing speed from multiple locations from multiple providers. I did give CacheFly example above but that is "not all" I am seeing. While the port does seem to be 1Gbps ... The results I am seeing very a lot.

That is why I even asked if someone know about more test files that are sure to be connected to 1Gbps port. (because it also depends on the other end what speed I am getting)

Umair

@bsdguy said:
How on earth could Umair have guessed that?!

@bsdguy said:
Oh, I got that. But shouldn't an expert like him know that good as well as bad dedis and VPSs can be at good or less good providers who again can be in good or less good colos and have good or less good connectivity?

not sure what to say to that ... I was never pissed at anyone or you. (But your replies seem like a different story)

You seem to have taken my comment above ("assuming lots ...") totally in a different way.
I am sorry if any of my comment made it look like i was pissed at you.

Umair

@rm_ said:
His point is that his brand new shiny expensive dedi is overall crappier network-wise than some $5 VPS from another provider, and that's not how it should be (or at least not what was expected).

Well, at least you see my point

I know it's not apple to apple thing ... A VPS could have 1Gbps shared port and a Dedi could come with 100Mbps unmetered .. etc etc.

But the over all result I am getting are kinda disappointing. And I did pay "extra" to get that 1Gbps port compare to their original offer. (And yes, I know it's not a dedicated port)

Anyway, I think we are going in whole different direction here.
All I was asking for is to have more test files (from providers with 1Gbps connectivity) so I can test speed from multiple locations ... And can have more data to process. I gave one example above (CacheyFly) which was just an indication (for me at least) that network is not that good. (When a provider advertise Premium network, I do expect it to be premium)

So back to topic:

Can you provide me some test files that I can try downloading and check speed test. (Don't have to be your own servers, but may be your provider have some. Looking for those providers who do have 1Gbps connectivity)

wych

http://ovh.net/ | http://mirror.dacentec.com/