> On Thu, 10 Jul 1997, HAL 9000 #3 wrote:
>
> > > I was too until someone brought up the lack of processor utilization
> > > reporting in ComOS, meaning there is no way to determine how hard you are
> > > hitting your CPU. Take a couple of full views, and do some filtering;
> >
> > I love the whole concept of CPU usage reporting. It's a perfect
> > example of Heisenberg's Uncertainty Principle. How much CPU was being
> > used before the CPU usage monitor was run? Who knows? It changes the
> > condition of the CPU by observing the CPU's condition.
>
> Personally, I feel much safer adding the overhead of a CPU usage reporting
> utility and knowing that I am *not* at the limit of what a given box can
> do with the reporting enabled. What is option #2? Waiting until I
> experience some sort of problem and just guessing, running the risk of
> negatively affecting the stability of a network, and possibly wasting time
> exhausting other possibilities until I find the source of the problem.
> Option #2(the current Livingston option) stinks(but I still like you guys
> for most everything else....)
>
>
On Thu, 10 Jul 1997 MegaZone wrote:
> Once upon a time Patrick Greenwell shaped the electrons to say...
> >Yes you have. So does engineering just kinda poke at the test boxes in the
> >lab, scratch their heads, and guess when a box is at capacity?
>
> No, we have external scopes we can hook onto the chip leads and see what
> kind of I/O is being done, how the bus is doing, etc. But it is an
> *external* monitor, non-intrusive, out-of-band. I don't see a good way
> include an ocilliscope in the unit. :-)
>
On Thu, 10 Jul 1997 Robert Hanson wrote:
> and until a solution is provided that allows one to know how hard their
> irx is being pounded with 2 full ds1's running bgp, many irx's will be
> used as low load non mission critical routers or as doorstops...
>
> -rh
>
> On Thu, 10 Jul 1997, MegaZone wrote:
>
> > I've explained many times now why CPU monitoring on Livingston units is
> > absolutely meaningless. The kernel design is not suited to monitoring.
> > I strongly doubt we will rewrite ComOS just to allow CPU monitoring since
> > it would be a major, low-level rewrite.
> >
>
Is'nt everyone saying the same thing here: If the box does not have
"blinky lights" (LEDs) or perf meter (Suntools) or some other indicators
that a symbolic processor (ie, Human being) can understand - then they do
not (and will not) get a warm/fuzzy that the "black box" is working
correctly and within its technical/engineering/configurable limits of
operation. After all, we all understand that every box has limitations!
The CPU meter, blinky LEDs, load line, Packets Per Second (PPS) etc.
simply provide basic information that lets a user know that the box is
performing correctly and within its operational limitations. So (MZ) if
you don't like the CPU meter, or the PPS meter or the blinky LEDs; please
suggest something else! Or put sufficient CPU/memory etc. resources
within the box that despite the most outrageous configuration, routing,
filtering, number of routes etc, etc., this box will simply not run out of
- well anything!! And _then_ we don't need any indicators.
This is why the Cisco 2501 is _everywhere_. I'm not saying its the best
router ever built, or that I love the modeful interface, or that I can
remember the crypic command set for one week to the other (I can't - very
casual user) - all I'm saying is that the operational limitations of this
box are very well understood and its very easy to use it _successfully_.
The IRX is at the other end of the usability spectum. Notice how many
users on this list say "...its the best thing since sliced bread..." and
other users say "...leave all my critical routing tasks to my Cisco...".
And do you think that tech support will tell the user - "well we don't do
this particular (routing) task well.... Why don't you buy a Cisco and stop
calling me." Is the IRX a deficient product? No. Does it have
limitations? Yes. Can you determine those operation limitations by
observation or by calling Tech Support? No. So why take a chance?
Did'nt we just have this discussion about the merits of front panel LEDs
on equipment and how the human brain can discern an unbelievable amount of
useful information from a seemingly random pattern of blinking LEDs.
Just my $0.02.
Al Hopper Logical Approach Inc, Plano, TX. al@logical-approach.com
(972)-379-2133 or (972)-849-5765. Fax 972-379-2134