Network slower than expected, how do I detect the cause?
My network topology is structure.
Code:
2 WAN NetGear FVS124G --> Dell PowerConnect 2724 When I put this in place I was getting 60MB/s transfers from my laptop to server. (which is the speed of the HDs) Now I get about 15MB/s almost consistently over my entire network. I tried power cycling the entire array and I'm now getting 38MB/s through iperf still only 15MB/s with scp. With a 1 GigBit Router I would expect it to be much higher since with a 100Mbit line I was getting 9MB/s transfers. The Xen array I'm getting anywhere from 2-10MB/s, I know I have a problem there with the xen machines talking to each other. Things that have changed: I've reformatted the server into a pile of xen machines. All servers I've switched from DHCP (controlled by 172.29.0.2) to Static IP. Upgraded two of my four home computers to Fedora 9 from Fedora 8. So my questions: 1. on a GigBit network what speeds are other people getting? Was my 60MB/s just a fluke and I'm looking for nothing? 2. Where do I start looking? Currently I'm still rebuilding the DB on my Xen array so I can't shut it off at the moment to see if it's the cause. Can any one point me to some places to check? |
Have you checked if there are collisions on your switch? How about the NIC of the servers?
|
Quote:
Just for fun I tried something: ------------------------------------------------------------ Server listening on TCP port 5001 TCP window size: 0.08 MByte (default) ------------------------------------------------------------ [ 5] local 172.29.0.83 port 5001 connected with 172.29.0.83 port 58961 [ 5] 0.0-30.0 sec 8030 MBytes 268 MBytes/sec I hit the interface card from itself. I get 268 MBytes/sec. So how did that work. Did that send traffic to the router then back to the card or did that all run within the card itself? |
All times are GMT -5. The time now is 08:20 PM. |