[gentoo-user] Re: [OT] etho app rating 10/100 etc in megabytes

[Harry Putnam]

James <wireless@×××××××××××.com> writes:

> Harry Putnam <reader <at> newsguy.com> writes:
>
>
>
>> But still, when I'm trying to measure how much data is moving
>
> emerge bwmon,
>
> It measures across the ethernet ports, so adjust your test,
> according to what you want to measure, crossing the ethernet
> port on the target system.

First off.. thanks for the tips and help.

All I get from bwmon is a large mess of incomprehensible data ending in 

,----
| b7841000-b7881000 r-xp 00000000 03:05 6663       /lib/libncurses.so.5.7
| [...]
| b789e000-b78ba000 r-xp 00000000 03:05 7228       /lib/ld-2.12.2.so
| b78ba000-b78bb000 r-xp 00000000 00:00 0          [vdso]
| b78bb000-b78bc000 r--p 0001c000 03:05 7228       /lib/ld-2.12.2.so
| b78bc000-b78bd000 rw-p 0001d000 03:05 7228       /lib/ld-2.12.2.so
| bffae000-bffc4000 rw-p 00000000 00:00 0          [stack]
| Aborted
`----

And it has no man page whatsoever.

(it has a little help at bwmon -h)

But I recall using bwmon years ago so not sure whats happening that it
crashes for me.

>> and it seems quite slow for what is supposed to be a gigabyte network.
>
> Gigabit ethernet rarely runs full out constantly, something, (ram, cpu,
> interface, swith-latency....) mucks things up. Do not let your "copper"
> get to long either!

But that figures out to about 3-4 MB per second (assuming I did the
math right)

I said it was averaging about 230-237 MBytes per MINUTE , so giving it
a nice round 240 MB per MINUTE:

  240 / 60 = 4MB per second... and that figures out to:

  ( using this forumula: 1MBytes ps = 8000000 bits ps or 8 Mbits ps)

  4 * 8000000 = 32 Mbits ps

That is not counting packets going the other way of course, but isn't
an incoming speed of 32Mega bits per second what one might expect from
adapters capable of 100 mbps... (not gigabit (1000))


What you've shown below appears to show gigabit network between h4 and
h5.  Is that really to be expected?

>>                    gigabyte switch
>>                    |            |
>>                    |            |
>>      (192.168.0.9) h4          h5 (192.168.0.17)

If you show the top half of the diagram you snipped, you see that h4
h5 are aimed at a switch/router/firewall above, that is only 100mbps.
The gigaswitch has no address, so I'm wondering if traffic between h4
and h5 has to go up thru the 100Mbps router to communicate with each
other. 

I realized when I made the diagram that I was probably looking for
gigabit speeds where really only 100mbps was possible.

Take another look at the diagram (Knowing that h4 and h5 have there
default routes set to the netgear (100mbps) router.

Would it still be possible that h4 and h5 would communicate direct
thru the gigabit switch or would that traffic have to go up thru the
100 Mbps router above?

(Note that in the previous diagram I had mislabled (just a typo) the
gigabit switch as gigabyte switch)

                     internet
                        |
                        |
                        |
             (netgear router is lan `default route'  <= 10/100***** 
           NETGEAR ROUTER (inside address 192.168.0.20)
                  |     |     |
                  |     |     |
   (192.168.0.5) h1     |     h3 (192.168.0.7)
                        |
                        |
                 gigabit switch
                 |            |
                 |            |
   (192.168.0.9) h4          h5 (192.168.0.17)


>> But also if I should be expecting h4 h5 to be able to use GigaByte
>> transfer speeds.
>
> Some fraction say 50% is good, if it is copper, unless the systems
> are smoking "gaming" systems or of very high quality resources.

I keep having a sneaking feeling I'm making some horrible mistake in
the math, but wouldn't the speeds I posted (240 MegaBytes per min)
figure out to something like 3.2+ % of the rated 1000 Mbits.

(I really hope I haven't demonstrated idiocy levels of math)