Asfand said:
60 Megabytes / sec = 480 Megabits / sec.
So, with 100Mbps LAN, the LAN is the bottleneck.
With 1000Mbps LAN, the drive is the bottleneck, but you're still
carrying about 4 1/2 times more data.
Megabits not megabytes. You made that speed mistake earlier and came to
the wrong conclusios regarding bandwidth utilization.
I'm doing a file transfer as I write this and am getting 56 megabits per
second rate based on the read speed of the server and the write speed of
the hard drive at the client.
Fact is that ALL hard drives are much slower then their advertised 100,
133, 150 or 300 megabit per second transfer rate. The fastest barely
cracks 70 megabits per second sustained.
See
http://www20.tomshardware.com/2005/09/27/round/ for supporting data.
Therefore, a gigabit LAN doesn't do ANYTHING for performance since the
bottleneck is INSIDE the server or client
Check to see if you currently have noetwork contention and address that
problem. It might be as simple as swapping out an ethernet hub for an
ethernet switch.
I'd rather have the hard drive as the bottleneck.
BTW, I plan to do most of my file transfers from a RAID 1 setup, which
gives the fastest sequential read speed.
I hate to burst your balloon, but that's not correct. RAID 1 is no
faster than a single drive and in real-world applications may actually
be a little slower. RAID 0 provides a speed improvement of up to about
60% which pumps it up to about 100 megabits INTERNAL transfer rates.
The make and model of the drives being used in RAID 0 will impact the
speed of the array. A pair of 80gb WD SATA drives in RAID 0 will not
perform as well as a pair of 160gb WD PATA drives in RAID 0.
Physics is just working against you when you move from theory to real
world applications with fragmented files. Its dooubtful whether you'll
actually see a sustained transfer rate greater than 100 megabits per
second into our out of your computer except for very short bursts from a
drive buffer.