View previous topic :: View next topic |
Author |
Message |
thompsonmike Apprentice
Joined: 19 Jan 2004 Posts: 275 Location: Bath UK
|
Posted: Thu Apr 07, 2005 3:57 pm Post subject: Network Speed |
|
|
I have a little problem with my network. It is laid out like this
Internet Modem (ADSL)
|
|
Gentoo Router
|
|
Network Switch
|
|
Gentoo Desktop x 2 / Wireless Access Point / Webcam
Now the whole system only runs at 10Mb/s tops. I would expect it to run at 50 - 60 Mbs.
All the network cards in use are reporting Full Duplex, 100Mbs apart from the one going to the ADSL modem, which is reporting Half Duplex 100Mb/s. Could this be a dodgy cable, Network Card? It is a mix of Netgear PCI cards and Cisco cards.
I have disconnected the WAP, and network cards, but the speed is still 10Mb/s
Any ideas on how I could up the speed a little? All the systems have DMA on their Hard Drives.
NFS Seems to be a little more zippy, but toward the end of the transfer it tails off to around 700Kb/s.
Please Please has any one got any ideas? _________________ Thanks
Michael.. |
|
Back to top |
|
|
mens Guru
Joined: 27 Aug 2003 Posts: 392 Location: Belgium
|
Posted: Thu Apr 07, 2005 4:05 pm Post subject: |
|
|
Make a crosslink cable and check the connection between the different devices 2x2 to detect the faulty one. Are you sure the switch is not the problem? Is it a real switch or a switching hub?
Have you sniffed your network to see if there is no unexpected bandwidth being used? |
|
Back to top |
|
|
thompsonmike Apprentice
Joined: 19 Jan 2004 Posts: 275 Location: Bath UK
|
Posted: Thu Apr 07, 2005 4:25 pm Post subject: |
|
|
mens wrote: | Make a crosslink cable and check the connection between the different devices 2x2 to detect the faulty one. Are you sure the switch is not the problem? Is it a real switch or a switching hub?
Have you sniffed your network to see if there is no unexpected bandwidth being used? |
It is a Switch, Linksys EZXS88W. I have run IpTraf on the network, and there is no extranious traffic there. _________________ Thanks
Michael.. |
|
Back to top |
|
|
thompsonmike Apprentice
Joined: 19 Jan 2004 Posts: 275 Location: Bath UK
|
Posted: Thu Apr 07, 2005 7:30 pm Post subject: |
|
|
Code: |
Settings for eth0:
Supported ports: [ TP MII ]
Supported link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
Supports auto-negotiation: Yes
Advertised link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
Advertised auto-negotiation: Yes
Speed: 100Mb/s
Duplex: Full
Port: Twisted Pair
PHYAD: 15
Transceiver: internal
Auto-negotiation: on
Supports Wake-on: pumbags
Wake-on: ubg
SecureOn password: 00:00:00:00:00:00
Current message level: 0x000040c5 (16581)
Link detected: yes
|
What is Port: Twisted Pair?
The other cards on the networks look like this
Code: |
Settings for eth0:
Supported ports: [ MII ]
Supported link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
Supports auto-negotiation: Yes
Advertised link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
Advertised auto-negotiation: Yes
Speed: 100Mb/s
Duplex: Full
Port: MII
PHYAD: 1
Transceiver: externel
Auto-negotiation: on
Supports Wake-on: g
Wake-on: d
Link detected: yes
|
_________________ Thanks
Michael.. |
|
Back to top |
|
|
mens Guru
Joined: 27 Aug 2003 Posts: 392 Location: Belgium
|
Posted: Thu Apr 07, 2005 7:44 pm Post subject: |
|
|
Twisted Pair means an RJ45 port.
What transceiver are you using on the other cards which use MII? |
|
Back to top |
|
|
thompsonmike Apprentice
Joined: 19 Jan 2004 Posts: 275 Location: Bath UK
|
Posted: Thu Apr 07, 2005 8:09 pm Post subject: |
|
|
transceiver? I am unsure as to what you mean by this.
That card is a NVidia built in jobbie on the Motherboard. I am using the reversed Enginerred driver in a 2.6.11-r4 Gentoo Kernel.
I was using a 3COM in this machine, but the new kernel had a right fit when I scrapped Windows and moved Linux only, complained about "Invalid MAC Address in EEPROM" and refused to even look at the card. _________________ Thanks
Michael.. |
|
Back to top |
|
|
mens Guru
Joined: 27 Aug 2003 Posts: 392 Location: Belgium
|
Posted: Thu Apr 07, 2005 8:34 pm Post subject: |
|
|
According to your 2nd ethtool output, you are using the MII interface on your card which has an external transceiver. What card is that? fyi, a transceiver is a device to link two different network type. (like BNC to TP). |
|
Back to top |
|
|
thompsonmike Apprentice
Joined: 19 Jan 2004 Posts: 275 Location: Bath UK
|
Posted: Thu Apr 07, 2005 10:13 pm Post subject: |
|
|
thompsonmike wrote: | transceiver? I am unsure as to what you mean by this.
That card is a NVidia built in jobbie on the Motherboard. I am using the reversed Enginerred driver in a 2.6.11-r4 Gentoo Kernel.
|
NForce _________________ Thanks
Michael.. |
|
Back to top |
|
|
thompsonmike Apprentice
Joined: 19 Jan 2004 Posts: 275 Location: Bath UK
|
Posted: Fri Apr 08, 2005 11:34 am Post subject: |
|
|
Just changed that the Network card in the client for a NetGear, and still getting the same results. All transfers no matter what are going out at 10Mb/s
_________________ Thanks
Michael.. |
|
Back to top |
|
|
thompsonmike Apprentice
Joined: 19 Jan 2004 Posts: 275 Location: Bath UK
|
Posted: Fri Apr 08, 2005 12:05 pm Post subject: |
|
|
I now dont belive this is a network problem, I read in another thread that reiserfs is dodgy on some things, so I did a SCP transfer to another host on the network and got 18Mb/s. I tried then to SCP it to 127.0.0.1 and got the same speed.
Looks like it may be a local issue, not on related to the network!
Code: |
mike@pherkab mike $ scp '/home/mike/Desktop/portage-20050328.tar.bz2' 127.0.0.1:1.test
Password:
portage-20050328.tar.bz2 100% 18MB 9.1MB/s 00:02
mike@pherkab mike $ scp '/home/mike/Desktop/portage-20050328.tar.bz2' 192.168.1.1:1.test
Password:
portage-20050328.tar.bz2 100% 18MB 9.1MB/s 00:02
mike@pherkab mike $
|
_________________ Thanks
Michael.. |
|
Back to top |
|
|
MrUlterior Guru
Joined: 22 Mar 2005 Posts: 511 Location: Switzerland
|
Posted: Fri Apr 08, 2005 12:17 pm Post subject: |
|
|
Emerge iperf to test your network speed. and post the results _________________
Misanthropy 2.0 - enough hate to go around
|
|
Back to top |
|
|
thompsonmike Apprentice
Joined: 19 Jan 2004 Posts: 275 Location: Bath UK
|
Posted: Fri Apr 08, 2005 12:31 pm Post subject: |
|
|
Thats just odd.
On the server
root@polaris mike # iperf -s
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 85.3 KByte (default)
------------------------------------------------------------
[ 4] local 192.168.1.1 port 5001 connected with 192.168.1.14 port 34837
[ 4] 0.0-10.0 sec 111 MBytes 92.8 Mbits/sec
On the Client
mike@pherkab mike $ iperf -c 192.168.1.1
------------------------------------------------------------
Client connecting to 192.168.1.1, TCP port 5001
TCP window size: 16.0 KByte (default)
------------------------------------------------------------
[ 3] local 192.168.1.14 port 34837 connected with 192.168.1.1 port 5001
[ 3] 0.0-10.0 sec 111 MBytes 92.8 Mbits/sec
Everything else reports under 10Mb/s. Can any one explain why??
That transfer was done through a SMB share. _________________ Thanks
Michael.. |
|
Back to top |
|
|
MrUlterior Guru
Joined: 22 Mar 2005 Posts: 511 Location: Switzerland
|
Posted: Fri Apr 08, 2005 12:54 pm Post subject: |
|
|
thompsonmike wrote: |
Everything else reports under 10Mb/s. Can any one explain why??
That transfer was done through a SMB share. |
I wouldn't believe anything SMB says about performance
Better use rsync & iperf combination to measure network performance and achievable copy performance. _________________
Misanthropy 2.0 - enough hate to go around
|
|
Back to top |
|
|
MrUlterior Guru
Joined: 22 Mar 2005 Posts: 511 Location: Switzerland
|
Posted: Fri Apr 08, 2005 12:55 pm Post subject: |
|
|
Two other possible factors which will contribute towards bad copy performance is your hdd & filesystem. Test both sides with hdparm. _________________
Misanthropy 2.0 - enough hate to go around
|
|
Back to top |
|
|
Shazam Apprentice
Joined: 23 Nov 2004 Posts: 191 Location: Germany
|
Posted: Sat Apr 16, 2005 4:26 pm Post subject: |
|
|
not to make matters worse, but isn't it a 100 Mbit card and you are taking of Mbytes being transfered ? that's a difference by a factor of eight. 100Mbits / 8 = 12.2 Mbytes. So you should be already pretty happy w/ your 10 Mbyte/s if that's the case, most of the rest probably falls for the overhead. |
|
Back to top |
|
|
thompsonmike Apprentice
Joined: 19 Jan 2004 Posts: 275 Location: Bath UK
|
Posted: Sat Apr 16, 2005 8:42 pm Post subject: |
|
|
Shazam wrote: | not to make matters worse, but isn't it a 100 Mbit card and you are taking of Mbytes being transfered ? that's a difference by a factor of eight. 100Mbits / 8 = 12.2 Mbytes. So you should be already pretty happy w/ your 10 Mbyte/s if that's the case, most of the rest probably falls for the overhead. |
Thats what I was looking for! Thanks for clearing that one up! _________________ Thanks
Michael.. |
|
Back to top |
|
|
WarMachine Apprentice
Joined: 15 Jul 2002 Posts: 181
|
Posted: Sun Apr 17, 2005 12:05 pm Post subject: |
|
|
I thought it was a 10 to 1 factor when considering network activity? |
|
Back to top |
|
|
Shazam Apprentice
Joined: 23 Nov 2004 Posts: 191 Location: Germany
|
Posted: Sun Apr 17, 2005 10:53 pm Post subject: |
|
|
well the ten to one factor can be consider that way, because you usally lose lots of bandwith to overhead. but the theoratic transformation comes from the bytes and bits. that one byte contains 8 bits ... |
|
Back to top |
|
|
|