Re: HUB slowing down network with only one user
From: Moe Trin (ibuprofin_at_painkiller.example.tld)
Date: 03/21/05
- Next message: Moe Trin: "Re: HUB slowing down network with only one user"
- Previous message: James Knott: "Re: Class B networks - number of networks/hosts"
- In reply to: Paul Wilhelm Elsinghorst: "Re: HUB slowing down network with only one user"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Sun, 20 Mar 2005 19:42:54 -0600
In article <pan.2005.03.20.10.33.50.605364@uni-bonn.de>,
Paul Wilhelm Elsinghorst wrote:
>Switching to a different port doesn't change the problem situation. But
>the device we are talking about is the sending one, not receiving. Still
>it works fine when using a crossed cable that's why I thought both nic's
>should be fine.
On a switch (at least every one I've ever had experience with) and many
powered hubs, each port has it's own "sender", This _could_ also be the
"receiver" on the port associated with the host that is sending the
packets. Does switching _that_ port make any difference?
>I'm still thinking the hub is somehow messing this up. I used to run it
>with two 10mbps nic's and it was all fine. Now that I switched to 100mbps
>I start getting problems.
Which brings up another question. What happens if you switch back to 10mbs.
That would almost certainly eliminate cables as a cause, but I'm not sure
if it would also be a valid test of the hub.
>Despite all the talk here about high traffic or not :-) it looks that
>after streaming audio data for about 10 minutes, like 1.5GB in 10 minutes,
>the connection gets bad.
An Ethernet should be expected to provide a maximum useful transfer of
about 30 percent of the port speed. Thus, a 100mbs should not be sweating
at 750 Megabytes per minute. Go beyond that rate, and you'll start seeing
collisions or dropped packets usually on the 'transmit' row in the ifconfig
stats. On a fully switched network, you can often get much higher transfer
rates, but the hardware (normally the switch, but the NICs can become a
factor) may start limiting things.
>My thoughts of getting a new switch are getting more serious.
I tend to agree there.
Old guy
- Next message: Moe Trin: "Re: HUB slowing down network with only one user"
- Previous message: James Knott: "Re: Class B networks - number of networks/hosts"
- In reply to: Paul Wilhelm Elsinghorst: "Re: HUB slowing down network with only one user"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
