Network Speed

[dcabbar]

Hi,

Would it be possible to provide the URL of a download file hosted in your Northern Virginia campus so that I can clock the download speed?

Also, can you please tell what bandwidth providers you use there?

Thanks...

[elix]

The providers are (listed in order that I see them used) : Cogent, ***, nLayer, Tiscali, ServerCentral.

[charles]

http://download.powervps.com/uumap.tar.Z

[heavyharmonies]

This has me concerned.

I've been doing comparisons over the last few days, between the file above and a large file on my current host. I've done tests at different times of the day, both from my home and from my office.

The download speeds to the above link are only about 50% of the speeds I'm currently achieving with my existing shared reseller account.

Are there any existing network issues that might be causing slowdowns, or is this indicative of the normal network speed?

For comparison, the file I'm using to compare with the above is:

http://www.heavyharmonies.com/uumap.tar.Z

Thanks,

-Dan

[Revised to make the target download the same]

[StingRay]

I just tried both files myself from work.
Found that the powervps was a little more than half the speed. (98 vs 178)


Here is a partial tracert.
3 16 ms 19 ms 19 ms a01.telus-a.starcon.com [208.181.128.1]
4 16 ms 18 ms 19 ms i1-2948g-l3.telus-i.starcon.com [64.114.240.1]
5 20 ms 27 ms 22 ms 207.194.64.173
6 19 ms 18 ms 19 ms 204.239.129.101
7 23 ms 19 ms 23 ms nwmrbc01gr01.bb.telus.com [204.174.217.37]
8 64 ms 69 ms 78 ms chcgildtgr00.bb.telus.com [154.11.11.18]
9 63 ms 66 ms 66 ms 154.11.3.214
10 99 ms 101 ms 117 ms pos4-1.br02.ash01.pccw***.net [63.218.44.50]
11 179 ms 132 ms 89 ms 63.218.94.94
12 87 ms 93 ms 148 ms eqashvzb2.defenderhosting.com [204.157.0.90]
13 86 ms 89 ms 89 ms download.powervps.com [204.157.9.117]

[heavyharmonies]

Here's the traceroutes to the two from my office this morning:


To current host:


Tracing route to heavyharmonies.com [66.227.8.90]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 128.174.64.1
2 <1 ms <1 ms <1 ms t-core2o.gw.uiuc.edu [130.126.0.233]
3 <1 ms <1 ms <1 ms t-core1o.gw.uiuc.edu [130.126.0.221]
4 * * * Request timed out.
5 <1 ms <1 ms <1 ms 130.126.0.14
6 <1 ms <1 ms <1 ms 130.126.0.30
7 <1 ms 2 ms <1 ms t-dmzo.gw.uiuc.edu [130.126.0.70]
8 <1 ms 2 ms <1 ms t-dmzb.gw.uiuc.edu [130.126.0.86]
9 26 ms 1 ms <1 ms 192.17.10.30
10 1 ms <1 ms <1 ms GE-6-2-2-100-7-Iron-CHM-P2-7-Babylon-CHM-Colo.lincon.net [206.166.9.73]
11 5 ms 5 ms 5 ms POS-3-0-1-NapSOB2-CHI-P2-7-Iron-CHM-P2.lincon.net [206.166.5.94]
12 5 ms 5 ms 5 ms pos-6-0-nap-sob2-nap-sob1.chicago.lincon.net [206.166.9.121]
13 5 ms 5 ms 5 ms ge-10-0-sob1-ans.chicago.lincon.net [206.166.9.6]
14 5 ms 5 ms 6 ms chcgil1wcx1-gige14-0-0.wcg.net [64.200.236.137]
15 5 ms 5 ms 5 ms brvwil1wcx2-pos5-0-oc48.wcg.net [64.200.103.57]
16 25 ms 25 ms 26 ms dllstx1wcx2-pos12-0-oc192.wcg.net [64.200.210.42]
17 25 ms 25 ms 25 ms dllstx9lce1-pos4-0-oc48.wcg.net [64.200.110.82]
18 28 ms 26 ms 26 ms dllstx9lce1-yipes.wcg.net [64.200.226.118]
19 29 ms 29 ms 29 ms 66.7.164.66
20 27 ms 28 ms 27 ms sancy.rwhmax.com [66.227.8.90]

To Powervps:


Tracing route to download.powervps.com [204.157.9.117]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 128.174.64.1
2 <1 ms <1 ms <1 ms t-core4o.gw.uiuc.edu [130.126.0.213]
3 <1 ms <1 ms <1 ms t-core1o.gw.uiuc.edu [130.126.0.237]
4 * * * Request timed out.
5 <1 ms <1 ms <1 ms 130.126.0.14
6 <1 ms <1 ms <1 ms 130.126.0.30
7 <1 ms <1 ms <1 ms t-dmzo.gw.uiuc.edu [130.126.0.70]
8 1 ms <1 ms <1 ms 209-254-131-77.ip.mcleodusa.net [209.254.131.77]
9 3 ms 3 ms 4 ms CHCGILWUH53JC01-SO2-1-0-0.mcleodusa.net [64.198.100.85]
10 3 ms 3 ms 3 ms CHCGILDTHBUJC01-SO0-3-0-0.mcleodusa.net [64.198.100.46]
11 5 ms 3 ms 3 ms ge3-0.cr01.chc01.pccw***.net [206.223.119.48]
12 76 ms 39 ms 37 ms pos4-1.br02.ash01.pccw***.net [63.218.44.50]
13 107 ms 100 ms 48 ms 63.218.94.94
14 31 ms 30 ms 31 ms eqashvzb2.defenderhosting.com [204.157.0.90]
15 33 ms 31 ms 30 ms download.powervps.com [204.157.9.117]


So even though the number of hops to PowerVPS is fewer, the throughput is lower. :(

[FrancescoV]

This is mine from italy :D
|------------------------------------------------------------------------------------------|
| WinMTR statistics |
| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| i.love.monica.bellucci - 3 | 142 | 138 | 187 | 199 | 234 | 203 |
| well.nicole.kidman.too - 2 | 142 | 140 | 187 | 199 | 250 | 203 |
| r-ba35-ba47.opb.interbusiness.it - 3 | 142 | 138 | 187 | 199 | 219 | 203 |
| host129-8.pool8020.interbusiness.it - 3 | 142 | 138 | 187 | 214 | 391 | 297 |
| r-mi256-rm215.opb.interbusiness.it - 1 | 142 | 140 | 203 | 223 | 375 | 235 |
| 151.99.75.226 - 3 | 141 | 138 | 203 | 219 | 359 | 219 |
| gw-mix-mi257-a.opb.interbusiness.it - 0 | 141 | 141 | 203 | 222 | 250 | 219 |
| ge-4-0-0.mil10.ip.tiscali.net - 1 | 141 | 140 | 218 | 241 | 297 | 250 |
| so-0-0-0.was10.ip.tiscali.net - 30 | 141 | 99 | 296 | 309 | 328 | 312 |
| dvlabs.cirn.net - 3 | 141 | 137 | 312 | 333 | 484 | 344 |
| corea.ash.dvlabs.com - 4 | 141 | 136 | 297 | 337 | 688 | 328 |
| eqash25.defenderhosting.com - 3 | 141 | 138 | 312 | 330 | 391 | 344 |
|unknownxxx.xxx.xxx.xx.defenderhosting.com - 3 | 141 | 137 | 296 | 326 | 359 | 328 |
|________________________________________________| ______|______|______|______|______|______|
WinMTR - 0.8. Copyleft @2000-2002 Vasile Laurentiu Stanimir ( stanimir@cr.nivis.com )

edit: actually i don't know if is an issue of my adsl provider or other things...

[Zaf]

I've been doing comparisons over the last few days, between the file above and a large file on my current host. I've done tests at different times of the day, both from my home and from my office.
The download speeds to the above link are only about 50% of the speeds I'm currently achieving with my existing shared reseller account.
Are there any existing network issues that might be causing slowdowns, or is this indicative of the normal network speed?
For comparison, the file I'm using to compare with the above is:
http://www.heavyharmonies.com/HRV_Sample.isoSorry, I did not understand the basis of your comparision in the first place. You are comparing download speeds of two networks using two different files. Why dont you upload the file provided by charles to your server and then compare it? Please note, I am not passing a judgement that PVPS network is better than your current host...but I suggest you use the same file and also make sure more you check from more than 4-5 locations/ISPs.

[heavyharmonies]

Actually, shortly after I posted that, I did upload the same file to my current host:

http://www.heavyharmonies.com/uumap.tar.Z

Gack! It's even worse now. :( :(

From the PowerVPS link, 89.8K/sec.

From my current host, 1.05MB/sec.

That's less than 1/10th the performance. :mad:

What are you seeing from your end?

Thanks.

[Zaf]

From PVPS:
[root@xxx root]# wget http://download.powervps.com/uumap.tar.Z
--15:53:08-- http://download.powervps.com/uumap.tar.Z
=> `uumap.tar.Z'
Resolving download.powervps.com... done.
Connecting to download.powervps.com[204.157.9.117]:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 8,520,221 [application/x-compress]

100%[====================================>] 8,520,221 1.25M/s ETA 00:00

15:53:15 (1.25 MB/s) - `uumap.tar.Z' saved [8520221/8520221]

From your server
[root@xxx root]# wget http://www.heavyharmonies.com/uumap.tar.Z
--15:53:35-- http://www.heavyharmonies.com/uumap.tar.Z
=> `uumap.tar.Z.1'
Resolving www.heavyharmonies.com... done.
Connecting to www.heavyharmonies.com[66.227.8.90]:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 8,520,221 [application/x-tar]

100%[====================================>] 8,520,221 892.76K/s ETA 00:00

15:53:45 (892.76 KB/s) - `uumap.tar.Z.1' saved [8520221/8520221]

[elix]

Here's the traceroutes to the two from my office this morning:


To current host:


Tracing route to heavyharmonies.com [66.227.8.90]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 128.174.64.1
2 <1 ms <1 ms <1 ms t-core2o.gw.uiuc.edu [130.126.0.233]
3 <1 ms <1 ms <1 ms t-core1o.gw.uiuc.edu [130.126.0.221]
4 * * * Request timed out.
5 <1 ms <1 ms <1 ms 130.126.0.14
6 <1 ms <1 ms <1 ms 130.126.0.30
7 <1 ms 2 ms <1 ms t-dmzo.gw.uiuc.edu [130.126.0.70]
8 <1 ms 2 ms <1 ms t-dmzb.gw.uiuc.edu [130.126.0.86]
9 26 ms 1 ms <1 ms 192.17.10.30
10 1 ms <1 ms <1 ms GE-6-2-2-100-7-Iron-CHM-P2-7-Babylon-CHM-Colo.lincon.net [206.166.9.73]
11 5 ms 5 ms 5 ms POS-3-0-1-NapSOB2-CHI-P2-7-Iron-CHM-P2.lincon.net [206.166.5.94]
12 5 ms 5 ms 5 ms pos-6-0-nap-sob2-nap-sob1.chicago.lincon.net [206.166.9.121]
13 5 ms 5 ms 5 ms ge-10-0-sob1-ans.chicago.lincon.net [206.166.9.6]
14 5 ms 5 ms 6 ms chcgil1wcx1-gige14-0-0.wcg.net [64.200.236.137]
15 5 ms 5 ms 5 ms brvwil1wcx2-pos5-0-oc48.wcg.net [64.200.103.57]
16 25 ms 25 ms 26 ms dllstx1wcx2-pos12-0-oc192.wcg.net [64.200.210.42]
17 25 ms 25 ms 25 ms dllstx9lce1-pos4-0-oc48.wcg.net [64.200.110.82]
18 28 ms 26 ms 26 ms dllstx9lce1-yipes.wcg.net [64.200.226.118]
19 29 ms 29 ms 29 ms 66.7.164.66
20 27 ms 28 ms 27 ms sancy.rwhmax.com [66.227.8.90]

To Powervps:


Tracing route to download.powervps.com [204.157.9.117]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 128.174.64.1
2 <1 ms <1 ms <1 ms t-core4o.gw.uiuc.edu [130.126.0.213]
3 <1 ms <1 ms <1 ms t-core1o.gw.uiuc.edu [130.126.0.237]
4 * * * Request timed out.
5 <1 ms <1 ms <1 ms 130.126.0.14
6 <1 ms <1 ms <1 ms 130.126.0.30
7 <1 ms <1 ms <1 ms t-dmzo.gw.uiuc.edu [130.126.0.70]
8 1 ms <1 ms <1 ms 209-254-131-77.ip.mcleodusa.net [209.254.131.77]
9 3 ms 3 ms 4 ms CHCGILWUH53JC01-SO2-1-0-0.mcleodusa.net [64.198.100.85]
10 3 ms 3 ms 3 ms CHCGILDTHBUJC01-SO0-3-0-0.mcleodusa.net [64.198.100.46]
11 5 ms 3 ms 3 ms ge3-0.cr01.chc01.pccw***.net [206.223.119.48]
12 76 ms 39 ms 37 ms pos4-1.br02.ash01.pccw***.net [63.218.44.50]
13 107 ms 100 ms 48 ms 63.218.94.94
14 31 ms 30 ms 31 ms eqashvzb2.defenderhosting.com [204.157.0.90]
15 33 ms 31 ms 30 ms download.powervps.com [204.157.9.117]


So even though the number of hops to PowerVPS is fewer, the throughput is lower. :(
Routers don't take ICMP packets with high priority, you should really ignore the ping times shown there. The main one you're looking at is the one at the end which is only around 30ms. Seems just fine really.

[elix]

Also, you'll find the following...

outgoing routes (traces from the VPS, this matters more for actual download speed) = Mostly Cogent, some (very little) DVLabs (DVLabs is nLayer, Tiscali [through Carpathia], NAC [supposedly], ***, and ServerCentral.

incoming routes (traces to the VPS) = mostly DVlabs and some Cogent. If you trace from Cogent's Looking Glass you actually see *** which is interesting.

There is only one reason the network could be setup like this...which is to save $.

Oh well, it's not really a big deal to me, but I'm sure it does have negative performance affects [perhaps not adverse ones, but they are still there]

Note: this is just from my observations...may not be fact.

[Zaf]

Note: this is just from my observations...may not be fact.Ditto observation about different routes for incoming and outgoing and always wondered why the difference. If its indeed just $$ v/s better performance.....dont blame me if I would select the latter. The same would apply if I were on the other side in the deal.

[albatroz]

If a customer asks you, what is the link speed of your VPS,
what do you respond?

[Robert]

The correct answer is 100MB :)

[Zaf]

I had seen a great deal of increase in 'failcnt' of the 'tcpsndbuf' parameter of user_beancounters more than a year back.
The support had incremented the softlimit of that parameter by 25% or so which stood at approx 6.5Mbps after the change.

[Robert]

Those paramaters all go away with SLM. :D