On Fri, 2002-10-11 at 06:27, Pierre Fortin wrote:
> On 10 Oct 2002 20:29:32 -0700 Jack Coates <[EMAIL PROTECTED]> wrote:
> 
> Jack,
> 
> Guess I've been quiet on this list for too long and don't recognize all
> the names; worse, not all new members know me...  oh well...  :>
> 
> First chill out...  my intent was to get the "me too'ers" to back off from
> using bing cuz it's a "cool tool"...
> 
> > On Thu, 2002-10-10 at 19:46, Pierre Fortin wrote:
> > <snip> 
> > > <flame>
> > > 
> > > Folks,
>     ^^^^^  == NOT targeted at an individual
> 
> The MAIN issue I was addressing was that people see some "neat _tool_",
> rush out and get a copy, then proceed to ADD to the problem(s) someone
> else is trying to resolve -- the equivalent of Internet
> "rubber-neckers"...
> 
> > First, you assume that I don't know the current user or bandwidth load
> > of the circuit, when in fact I know both and am working with my ISP to
> > troubleshoot a problem.
> 
> "working with my ISP" is NOT clear proof that either or both parties
> understand the symptoms, the effects, etc...  worse, _altering_ the
> problem by injecting additional traffic which changes what is being
> analyzed.  Increasing traffic this way will only move you closer to the
> "avalanche" mode quicker -- it will not help you figure out the problem
> (unless you are looking for a one in which all-0s or all-1s is impacting
> the link).
> 
> > <snip> 
> > > "Tools" like this are unscientific and in an indirect way a DDoS
> > > attack on the network.
> > 
> > And second, you assume that bing is in the same class as rape.c. From
> > the highly edifying man page:
> > DESCRIPTION
> >        Bing determines bandwidth  on  a  point-to-point  link  by
> >        sending  ICMP  ECHO_REQUEST  packets  and  measuring their
> >        roundtrip times for different packet sizes on each end  of
> >        the link.
> > 
> > The actual packet stream used to measure bandwidth on the DS-3 was 3.6
> > kbps.
> 
> So, can you tell me why it would even be necessary to measure this
> link...?  A DS-3 is (rounded) 45Mbps.  If this is a *fractional* DS-3, I
> could see the interest in finding the speed; but I'd simply *ask* the ISP
> who **should** know how it's deployed -- especially when "working with my
> ISP"

Anyone who thinks that an ISP knows what they are doing just because
they are an ISP isn't up on on the ins and outs of the ball game. 

> 
> Besides, since you haven't found & fixed the problem, you can't say that
> the link is not just 2kb from avalanching...
> 
> Do you know the characteristics of all the components involved in "round
> trip times"...?
> Here's a *partial* list:
> - available memory (sender, routers, target)
> - queue sizes (tx & rx)
> - buffer availability (tx & rx)
> - prioritization (sender, routers, target)
> - queue position ( " )
> - traffic already on wire
> - bit transit time
> - bit insertion (gapped clock): time on wire is longer than packet size
> - packet size (changes RTT -- duh)
> - packet loss
> - retransmission
> - "avalanching"
> - accuracy of analysis "tool"
> - etc....
> 
> Generally, over the years, I've found this type of "tool" to be more of a
> hindrance in finding the root cause of problems...  I can't recall the
> number of times I've found the "tools" to be bogus...
> 
> > Thanks for playing!
> 
> When it come to problems, I don't play...  usually, I end up being called
> upon because there's more than one cause which the average troubleshooter
> can't figure out...  I don't know what's going on in your case; but I'd
> bet you that bing will just delay finding the real cause(s)...
> 
> Tell us your problem(s)** and maybe we can help...  Sorry, I didn't follow
> this entire thread -- IIRC, the original poster wanted to know how to
> determine 10 vs 100 Mbps ethernet...  Using tools like bing on a locally
> attached, personal link is one thing; using it on shared 'net resources is
> an abuse, IMO -- and the reason for my flame.
> 
> ** just the facts...  can't begin to tell you the number of problems where
> I've had to debug the reporter's analysis/interpretation *before* I could
> begin to debug the actual problem...
> 
> Hope this makes sense... I just woke up and am still groggy...  :^)
> 
> Enjoy,
> Pierre

Pierre back down Please.  This is a neat tool yes.  It does have it's
place yes.  It doesn't add any more traffic than normal pings.  Take a
look at the tool and it's algorithms.  You'll find that the numbers it
produces are only accurate on a Lan but over a connection from say an
office to the local Telco, level 3 etc etc.  It can be very useful for
providing proof that Your Unix box isn't the reason you've been
experiencing a 40% drop in traffic speed over the last 3 months.  (Yes
Level 3 and Pac Hell have both told me that our Linux Firewalls and
servers were the reason that our t-3 had intermittent speed drops of 75%
or more at really unusual times.) As if running M$ crap would suddenly
make a bad fiber splice good again. (Bing proved the slowdown.  and an
OTDR showed the location of the break.)  

James

> 
> ----
> 

> Want to buy your Pack or Services from MandrakeSoft? 
> Go to http://www.mandrakestore.com



Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com

Reply via email to