Gus Wirth wrote:
If I do an NFS mount and copy a large file from the server to /dev/null on the client or if I scp a file between the server and client I get pretty much the same results, about 11MB (that's bytes) per second. This is only about 10% better than 100baseT ethernet. I know from previous experiments that my hard drives and general system throughput can handle about 25MB/sec.
Many cheap gigabit switches aren't wirespeed. Also, beware the 32 bit, 33MHz PCI woes. 32/33 PCI is only capable of 132MByte/sec at maximum. Also keep in mind that other things are happening on PCI as you're doing the transfer, namingly slamming the hard disks, etc.
So any hints on how to get this to work faster, or am I seeing the results of a marketing scam?
Try a better switch. You may not need this many ports, but usually things with 24+ ports are better than consumer grade and offer better throughput. Also, try a better board, something with the intel E1000 chipset or broadcom chipsets. also good is alteon acenic.
to isolate the switch as a potential choke point, connect a garden variety crossover (cat 6!) cable between the two machines and set IP addresses manually, etc., then start the copy. see if performance increases. also, make sure all your cables are cat6. cat5 wasn't meant for gigabit speeds. cat5e may be able to handle them but I think it's only rated to atm over copper speeds (155MBit/sec). so yeah. make sure you use new cables. generally, the machine-built ones are better than the hand-built ones, since the machine can align the wires better without removing twists.
if you ask me, there's only one way to do gigabit and get any proper performance/reliability out of it: 1000Base-SX; that is, gigabit over multimode fiber, and high-end equipment (boards and switches). at my house i have a cisco 3550 switch which has two GBIC ports; connected to these are SX GBICs -- one goes to my workstation which has one of the aforementioned AceNICs, the other will uplink to a second switch someone at work is giving me.
good luck! -kelsey -- [email protected] http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
