----- Original Message -----
Sent: Thursday, May 02, 2002 12:10
PM
Subject: [Bridge] question/discussion
about performance of the bridge
Hello,
Can someone tell me if I am correct so far?
I am trying to figure out if my bridge (cpu) can handle my theoretical maximum
throughput.
I apologize, been up 20 hours working on this
bridge (very exciting) and am not sure at this point if 1kbit = 1000 or 1024
or 1048, but I believe 1000 is close enough for now.
My ISP provides a service to me which has a
downstream throughput of 768k bit/s = 768,000 bits/s. A
packet = 1500 bytes = 12000 bits, based on my
understanding of MTU. If I divide 768,000
bits/s by 12000 bits I should have a theoretical maximum throughput of 64
packets/second over this bridge. I assume all packets are always 1500
bytes? I assume latency is out of the
picture and I have a fully flowing pipe. I have the bridge attached
between a decent switch and a DSL modem, so I assume there is no overhead due
to collision. The way I understand it,
each packet, hitting each NIC generates a CPU interrupt, so two NICs in bridge = 128 interrupts per second (not sure if
packet hitting outgoing NIC generates an interrupt). Oops, forgot, Then
I add the upstream is 128kbits or 15% more = ~148 interrupts per sec during
theoretical maximum throughput.
Can someone give me a start on how to figure out
the amount of the above interrupts a processor of varied size can
handle? I am thinking a cpu can, in
theory, handle a certain amount of interrupts per MHz.
Is nettimer the best tool for finding a
Intranet/Internet bottleneck?
Regards,
Brent
Seagrave
