[yonatan]

I see no problem with the volume network and it works well. I do see problems with the "mirror.leaseweb.com" domain outside of the LW network though.

I had to migrate an SQL db from a "volume" server to ovh ssd range, it downloaded the DB at 10Mbit max , and it was over to an EG server.

what hw was you using there?

[_Lemon_]

Does OVH actually discriminate the quality of bandwidth each type of server range receives?

Yes.

Do NOT use Leaseweb as any form of calculation. Most of their high traffic mirrors and volume customers are on what they call the Volume Network which is horrible (I use one on the Vol Net)... not because of saturation, but because there's only one point you connect through - it's volume because it's cheap/not multi-path. Their Standard Network is decent, but they don't put their pubs on that.

I see no problem with the volume network and it works well. I do see problems with the "mirror.leaseweb.com" domain outside of the LW network though.

[Tz-OVH]

that's good , if the driver helped, its the drivers fault...
its not the same hardware after all :-)

It's way way better hardware. The PRO server's CPU is C2D E6550 vs Kimsufi's i7-950, 2GB vs 12GB RAM, HDDs are faster on the new server as well.

[yonatan]

that's good , if the driver helped, its the drivers fault...
its not the same hardware after all :-)

[Tz-OVH]

so it's fixed now? the driver was it?

I think its fixed...but perhaps I'm also on a very heavy use rack. I notice quite a disparity during some single and multi-threaded speed tests from my PRO vs Kimsufi server.

[yonatan]

so it's fixed now? the driver was it?

[Tz-OVH]

what's your ethstatus data?
check for RX/TX error on the NIC, it could be because a cable defect.... or check for driver update for your NIC.

Had that problem with an SP-Mini.

The best thing you can do, is provide OVH support with traceroutes and ask them to check if its good.

I once had an NL download coming from London , instead of Direct from NL.

This might be the case ....

btw .. .why you test only download? , what about upload? is it slow on both streams?

Not sure how to check RX/TX errors on the NIC.

Testing upload

Old PRO server (100mbit)
http://img831.imageshack.us/img831/2548/100mbit.jpg

New Kimsufi server (gbit)
http://img524.imageshack.us/img524/419/gbitz.jpg

Updating the NIC drivers seems to have done wonders. I thought perhaps being a fresh install, Windows would have the latest drivers...

Same test file @ Leaseweb:

C:\>wget http://mirror.leaseweb.com/FreeBSD/I...86-dvd1.iso.gz
--2010-10-11 05:25:15-- http://mirror.leaseweb.com/FreeBSD/ISO-IMAGES-i386/8.1/
FreeBSD-8.1-RELEASE-i386-dvd1.iso.gz
Resolving mirror.leaseweb.com... 94.75.223.121
Connecting to mirror.leaseweb.com|94.75.223.121|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 1996298670 (1.9G) [application/octet-stream]
Saving to: `FreeBSD-8.1-RELEASE-i386-dvd1.iso.gz'

100%[====================================>] 1,996,298,670 14.4M/s in 1m 43s

2010-10-11 05:26:59 (18.4 MB/s) - `FreeBSD-8.1-RELEASE-i386-dvd1.iso.gz' saved [1996298670/1996298670]

[glidewave]

Do NOT use Leaseweb as any form of calculation. Most of their high traffic mirrors and volume customers are on what they call the Volume Network which is horrible (I use one on the Vol Net)... not because of saturation, but because there's only one point you connect through - it's volume because it's cheap/not multi-path. Their Standard Network is decent, but they don't put their pubs on that.

[Thelen]

Test via http://proof.ovh.net/ for a start not via LW which can be saturated at times.

Otherwise, yea maybe rack full of seedbox, but even then not that likely because the techs would see 100% utilisation and do something about it..

[DigitalDaz]

A speedtest result of a virtual machine on that server gives:



I too am on the 188.165.x.x range

Running Proxmox

The speedtest is on an Ubuntu Desktop KVM

[DigitalDaz]

I also have an i7-4T with the gigabit port:

Resolving mirror.leaseweb.com... 94.75.223.121, 2001:1af8:4030:1:215:17ff:fecf:3668
Connecting to mirror.leaseweb.com|94.75.223.121|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 1996298670 (1.9G) [application/octet-stream]
Saving to: `FreeBSD-8.1-RELEASE-i386-dvd1.iso.gz'

100%[====================================>] 1,996,298,670 18.8M/s in 69s

2010-10-11 00:54:35 (27.7 MB/s) - `FreeBSD-8.1-RELEASE-i386-dvd1.iso.gz' saved [1996298670/1996298670]

[yonatan]

what's your ethstatus data?
check for RX/TX error on the NIC, it could be because a cable defect.... or check for driver update for your NIC.

Had that problem with an SP-Mini.

The best thing you can do, is provide OVH support with traceroutes and ask them to check if its good.

I once had an NL download coming from London , instead of Direct from NL.

This might be the case ....

btw .. .why you test only download? , what about upload? is it slow on both streams?

[Tz-OVH]

I don't think so. Only unlimited traffic servers have different bandwidth quality (including RPSes).

It's more luck than anything else. You might end up on a rack full of seedboxes.

I've submitted a technical ticket...maybe they can tell me if there's a setting issue somewhere on their side.

I feel shortchanged though, I expected much better performance from this server, not substantially less! Considering just renting another SP Mini server

[Razakel]

Does OVH actually discriminate the quality of bandwidth each type of server range receives?

I don't think so. Only unlimited traffic servers have different bandwidth quality (including RPSes).

It's more luck than anything else. You might end up on a rack full of seedboxes.

[Tz-OVH]

I just rented an i7-4T, with the exact same software setup (Win 2008 R2 Standard x64). I noticed the last few days its not been as fast for a Gigabit server (I paid the Pro fees of 15GBP + Gigabit option 10GBP).

A simple test with an Ubuntu 10.10 release torrent showed me rather startling results.

SP-09 Mini, with 1minute I was downloading at full 100mbit
i7-4T, only at the very end of the download (average of 10mbit till the end), I managed to barely reach 50mbit.

Tried the same comparative tests with direct http/ftp downloads of various linux builds, same result. My older, 100mbit limited, PRO box was faster than my much newer 1gbit connection based Kimsufi box.

Does OVH actually discriminate the quality of bandwidth each type of server range receives?

Both servers are of the Traffic series. PRO server is on the 91.121.x.x range, Kimsufi is on the 188.165.x.x range, perhaps RBX3.

I'm seriously considering just returning the Kimsufi for a full refund and either sticking with my PRO server or upgrading to a better PRO server.

KIMSUFI server: (gigabit)

C:\>wget http://mirror.leaseweb.com/FreeBSD/I...86-dvd1.iso.gz
--2010-10-10 20:11:23-- http://mirror.leaseweb.com/FreeBSD/ISO-IMAGES-i386/8.1/
FreeBSD-8.1-RELEASE-i386-dvd1.iso.gz
Resolving mirror.leaseweb.com... 94.75.223.121
Connecting to mirror.leaseweb.com|94.75.223.121|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 1996298670 (1.9G) [application/octet-stream]
Saving to: `FreeBSD-8.1-RELEASE-i386-dvd1.iso.gz'

100%[====================================>] 1,996,298,670 6.25M/s in 5m 14s

2010-10-10 20:16:40 (6.06 MB/s) - `FreeBSD-8.1-RELEASE-i386-dvd1.iso.gz' saved [1996298670/1996298670]

PRO Server: (100mbit)

C:\>wget http://mirror.leaseweb.com/FreeBSD/I...86-dvd1.iso.gz
--2010-10-10 23:13:33-- http://mirror.leaseweb.com/FreeBSD/ISO-IMAGES-i386/8.1/
FreeBSD-8.1-RELEASE-i386-dvd1.iso.gz
Resolving mirror.leaseweb.com... 94.75.223.121
Connecting to mirror.leaseweb.com|94.75.223.121|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 1996298670 (1.9G) [application/octet-stream]
Saving to: `FreeBSD-8.1-RELEASE-i386-dvd1.iso.gz'

100%[====================================>] 1,996,298,670 7.53M/s in 4m 12s

2010-10-10 23:17:45 (7.56 MB/s) - `FreeBSD-8.1-RELEASE-i386-dvd1.iso.gz' saved [1996298670/1996298670]