Check is my arithmetic calculating lan throughput..

[jmille34]

I'm trying to take some benchmarks on network storage devices (specifically a buffalo terastation), so I'm copying huge directories onto it straight through from a w2k server to the nas box, no switch, using a trendnet gigabit card and a 3ft belkin cat6 cable. And I'm using perfmon to graph the throughput, but I don't know if I'm doing my calculations correctly.. it seems like I'm off somewhere.

Perfmon is averaging 3,000,000 (7 digits), and it says the unit of measurement is bytes/second. So if I take that, multiply by 8 to get bits/second, then divide by 1000 (or 1024) to get kilobits/second, and then divide again by 1000 (or 1024) to get megabits/second, I come up with 24mbps (or 22.88mbps).. Is this speed the same unit of measurement as the 1000mbps stamped on a gigabit network card as the maximum theoretical throughput?

If my numbers are correct, that seems awefully low. I realize there are other factors such as the hard drive speed and tcp/ip overhead, but I would think 70megabit would be reasonable even for a 100mpbs lan.

Any ideas/corrections?

 

[Seaspray0]

tcp/ip overhead can be quite impressive but not that much. On tests I ran once on a 10mb network, I was transfering about 1.5mb/sec of actual data.

A+/MCP/MCSE/MCDBA

 

[codered10]

I think your speed restriction is probably occuring due to issues like processor and hard drive speed.

Thanks a ton,
codered10

 

[jmille34]

Yeah, I tried it from a 10k rpm sata drive, and the rate was much higher.. like 122mpbs or thereabout. The pc I was using is an amd 2ghz with 2.5gigs of ram, but it turns out that the D: drive with the data is a total turd. I knew it was a seagate 7200.x, so I thought it was a sata drive, but it turns out it's an ata/100 drive.. cheap, but huge.. doh!

Anyway, I just wanted to make sure my arithmetic was right, though. I could have messed up a decimal or something so that it was running at 240mbps, which would have been fine.