🚀 go-pugleaf

In article <34610C44.4106@calgary.chevron.com>, 
dego_nospam@calgary.chevron.com says...
>
>seanc wrote:
>> It's not dedicated bandwidth, it is a shared broadcast medium, and when the
>> you approach the supposed ~37% max utilization you will begin to feel a 
severe
>> performance hit, I have witnessed this time and again on corporate net's so 
am
>> a believer. 
>
>Very often, if your LAN has problems at that low a utilization, there
>are network problems or design flaws which may be the cause (not the
>result) of those problems.  Not always is this the case, but I've seen
>it often enough not to rule these kinds of things out.


I agree completely, my point is that on an ether segement, regardless of
design(unless you are talking micro-segmentation, but that's another story) 
once utilization gets into the 35 to 40 percent range, collisions
increase dramatically, and performance will begin to degrade rapidly.

>
>> My average speed to my gateway is nearly 500K consistently when doing adhoc
>> testing. once it passes my gateway, bounces off two or three nodes and hits
>> the Net at large, my throughput is more like 1.5KBps to 50KBps.
>
>Using a P133/32MBram/Win95/3c905/WS_FTP on a Motorola CyberSURFR cable
>modem system to a server at the cable company, I've gotten up to 720
>KBps.  I don't know what this is in bps if you include overhead.  This
>was after I applied a patch to Win95 that I believe updated the window
>size.  Before this, I was only getting 220KBps.  I also changed the
>3C905 card setting to maximum throughput, which also uses your CPU more.
>
>Now, most sites on the net I get 20-50 KBps.  However, some mirrors such
>as Netscape or Microsoft give me over 500 KBps.  Not bad.
>
>(yes, all those numbers are BYTES per second, not bits)

of course

>
>> As someone else noted though, testing a live network is utterly pointless.
>
>I would not say that. Certainly its usefull to compare it to itself
>during various points in time.  When I first got my cable modem last
>Feb, I ran traceroutes, pings, and ftps to many sites, during the day
>and night, and recorded them all.   Many people commented that
>performance would be terrible soon as anyone else got on the cable modem
>network with me.  So this Feb I plan to run the same tests over again,
>night and day, and post both sets of results so people can see for
>themselves.

for baselining I will agree, it just gives you a picture.

>
>Certainly, there are some days when the Internet dies for one reason or
>another.  When I suspect there is something wrong, I start
>pinging/tracerouting to various sites, and comparing the results to the
>previous results when things are running good, you can see where the
>problems are originating.  If you are an ISP buying bandwidth from
>someone else, you do monitoring on a regular basis so that you can
>identify if your bandwidth provider has chronic bandwidth or routing
>problems that keep re-occuring.