RE: Every time ..., was Re: General formula

[Andrew Probert <AndrewP@xxxxxxxxxxxx>]

Two bobs worth ..

Erlang calculations have been tablulated to give people an easy lookup for
telephony calculations.  Given a presented load in seconds, and a number of
lines, what is the percentage of calls that will get a busy signal?

It works, given 3 factors you can lookup any way you like.

However for this, or any capacity planning exercise, real-world data and
assumptions are input to the model.  

Erlang tables require the user to input their assumptions on presented load
(an average) and number of lines avaliable to derive busy.  Alternatively
how much busy time they can tolerate, for a given load to derive a minimum
number of lines.

I'm all for a model, please send me the Excel spreadsheet implementation
when its complete.  

However, I think the real world of certs is too complex.


Andew Probert
Rotek Consulting   http://www.rotek.com.au
a Division of Secure Network Solutions
Tel  +61 3 9690 8877
Fax +61 3 9690 8171



> -----Original Message-----
> From:	Ed Gerck [SMTP:egerck@mcg.org.br]
> Sent:	Friday, June 04, 1999 6:09 AM
> To:	Ben Laurie
> Cc:	Alan Lloyd; Bob Blakley; Graham Klyne; PKIX
> Subject:	Re: Every time ..., was Re: General formula
> 
> 
> 
> Ben Laurie wrote:
> 
> > Ed Gerck wrote:
> > > This also means that if you call someone and it is busy, you could
> call
> > > right afterwards (just add the round-trip total delay time -- say, 5
> sec)
> > > and your sucess rate should be the same as if you would wait some
> > > minutes (as people normally do).
> > >
> > > The reason in both cases is simple -- since phone statistics is given
> by
> > > a Poison distribution and you don't know when the conversation
> actually
> > > *started*, its end does not depend on the duration of your
> observation.
> >
> > I don't believe this for a moment. To take an extreme example, if people
> > make, on average, 1 phone call a year, with a mean duration of 1 minute,
> > then it is intuitively obvious that if I wait a day after they are busy,
> > I have an almost 100% chance that they won't be busy,
> 
> Ben Laurie:
> 
> Is this how people really use the phone where you live? Is this "phone
> statistics"? C'mon, give me a break-- and realize that the statistics of
> your "extreme example" (what an understatement) is not even given by
> a Poisson pdf.
> 
> > whereas 5 seconds
> > later the chances are noticably less than 100%. Of course, I believe the
> > second statement, but the first is not a consequence of it, IMNSHO.
> >
> > You're just making this stuff up as you go along, aren't you?
> 
> You remind me of a guy that drowned in a lake that had just one inch
> of water depth, on average. Just too bad he did not realize it was an
> average, though. Or, even, what an average was.
> 
> Since phone traffic statistic is hardly on-topic here, I might help you
> off
> list if you consider this important for your understanding of certificate
> lifetimes.
> 
> And, Ben, next time you don't understand something, just ask and say what
> part of my text you did not understand. You don't need to become
> aggressive
> in order to receive my reply.
> 
> Cheers,
> 
> Ed Gerck
>