home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Software Vault: The Gold Collection
/
Software Vault - The Gold Collection (American Databankers) (1993).ISO
/
cdr35
/
showfr12.zip
/
README.DOC
< prev
next >
Wrap
Text File
|
1993-06-19
|
14KB
|
303 lines
System Availability Analysis - SHOWFREE
---------------------------------------
Copyright 1993 Clark Development Company, Inc.
Description
===========
The program contained herein, called SHOWFREE, is unlike many of the ordinary
system usage analysis programs available for PCBoard BBS's. Instead of trying
to analyze HOW MUCH your system is used (which is something that is important
for the sysop to know) it focuses instead on HOW FREE your system is (which is
something a potential caller would want to know).
Here is an example of the difference in thinking behind the two different
system analysis schemes:
Given the following: 3 nodes running in rotary on a single phone line
Node #1 is busy for an entire hour
Node #2 is busy for 30 minutes of this hour
Node #3 never gets a call this hour
How much system is used: Utilization is 90 minutes spread across 3 nodes.
Some analysis programs will show this as being
50% utilization because there were 180 system
minutes available but only 90 were used.
How free is the system: The system is 100 percent free!!! Because there
were NO calls that rolled down to node 3 (from
nodes 1 and 2) there was NEVER a time that a
caller would have received a busy signal. To
the caller - the system was 100% free that hour.
So while the system usage analysis showed 50% usage, something the sysop needs
to consider when trying to decide if his system is being properly utilized,
the system availability analysis instead showed that the system was 100%
available.
Here again, this data can be important to the sysop. If the system
availability drops below 50% then the sysop should realize that half the time
the system is up the callers are getting BUSY SIGNALS and a new phone line may
be needed.
Let's do one more example just to make sure we understand the difference:
Given the following: 3 nodes running in rotary on a single phone line
Node #1 is busy from 1:00 to 1:50
Node #2 is busy from 1:20 to 1:45
Node #3 is busy from 1:15 to 2:00
How much system is used: Utilization is 120 minutes spread across 3 nodes.
Some analysis programs will show this as being
67% utilization because there were 180 system
minutes available and 120 of them were used.
How free is the system: The system has 20 minutes, from 1:00 to 1:20,
when there is at least one node free. Then from
1:45 to 2:00 there are another 15 minutes free.
All told, there were 35 minutes out of the hour
when a caller could have connected without a
busy signal. That makes the system availability
58 percent for that hour.
Program Requirements
====================
This program will ONLY work properly given the following requirements:
- A given CALLER log contains at most 24 hours worth of entries
- If an event is identified to the system it is assumed that all nodes
in a group will be down for the same period of time for the event
- Multiple nodes on a single rotary or ring-down are grouped together
This program saves a history of system availability and averages everything
together. Therefore it does not need to have more than one day's worth of
information in the CALLER log and in fact putting more than one day's worth
of information into the file can cause errors in the calculations.
It is therefore recommended that you delete your CALLER logs once a day. If
you have other software that needs to use more than one day's worth of
information then you can append each day's log to the master log like this:
COPY \KEEP\CALLER1 + CALLER1 \KEEP\CALLER1
The above line would add CALLER1 onto the end of \KEEP\CALLER1.
Line Groups
===========
This program will work fine on a single node system. However, the analysis
of system availability is easily predictable as the reciprocal of a system
usage analysis. In other words, if the system is 25% used then it is 75%
free.
Where this program really shines is when determining the system availability
across multiple nodes in a single line group. By that, I mean, if you have
had Ma Bell set you up with 5 phone lines and 3 of them ring in rotary for
a set of HST modems, for example, and the other two ring in rotary for a set
of Hayes V-Series modems - then you have a very obvious separation of two
line groups. One for HSTs and another for V-Series.
To make this easier... if a caller can dial ONE PHONE NUMBER and reach more
than one node - then all of the nodes that he can reach are in the same group.
Program Installation and Usage
==============================
This program is pretty easy to install. Just make a \PCB\SHOW directory in
which to run the program in (any unique subdirectory will do). Copy all of
the files into that directory. Your EVENT.SYS file will then contain a
line such as this:
showfree showfree.cnf
The above line will create a file which you can then use as a bulletin on your
system. The format of the SHOWFREE.CNF is as follows:
------------------------------------------------------------------------------
C:\PCB\MAIN\CALLER # name/loc of caller logs
1-4 7-9 11 # node numbers to process
SHOWFREE.RPT # name of report file
SHOWFREE.SAV # name of average history file
02:00-02:30 # event start thru event end times
# everything below this line is a "header" for the report file
@CLS@@X0FPHONE: 801-261-8977 (nodes 1, 2, 3, 4, 7, 8, 9, 11)
@X0B
System Availability Graph
(chances of getting through on first call)
------------------------------------------------------------------------------
The first 2 lines of the above identify where the caller logs are to be found
and which node numbers to put into a group. You can list each node number
separately as in "1 2 3 4 5" or you can use a dash as in "1-5" to mean the
same thing. Spaces must be used as number separators. Commas cannot be used.
The next line defines the name of the report file (or bulletin if you want to
have the file immediately posted as a bulletin).
The next line defines the name of the history file to be used. If you are
using multiple groups then it is recommended that each group have its own
history file.
The next line defines the time of the event. It is not necessary to list an
event and in fact the line can be left blank. If an event is specified it is
assumed that all nodes within the group will be down for the period of time
shown for the event.
The seventh and all subsequent lines are the report header. You can put
anything you want and make the rest of the file as long or as wide as you
wish. The only rule is: make it readable for your callers.
Version 15.0 Additions
======================
PCBoard v15.0 offers the ability to monitor missed connections. There are
three possible types of "missed connections" that are reported in the callers
log. They are:
NO CONNECT/NO CARRIER - indicates a ring was detected, but the modem
reported NO CARRIER during the countdown
sequence before a connection was made
NO CONNECT/FULL COUNTDOWN - indicates a ring was detected, but that the
system counted all the way down without a
connection ever having been established
Carrier Lost - if carrier lost shows up in between calls
without a caller name to show logon/logoff then
this indicates that a connection was established
but that the caller lost carrier before typing
in a name.
In order to make use of these values, however, you MUST modify your PCBTEXT
file. The following commands can be used to make the modifications:
MKPCBTXT C:\PCB\GEN\PCBTEXT /I:34 "@SYSTIME@ NO CONNECT/@OPTEXT@"
MKPCBTXT C:\PCB\GEN\PCBTEXT /I:215 "@SYSTIME@ Carrier Lost ..."
These two changes put the TIME of the occurance into the log so that SHOWFREE
can track when the problem occur.
Finally, you must go to PCBSetup : Modem Information : Configuration Switches
and answer 'Y' to the question "Monitor Missed Connections" in order to enable
the above lines in your callers log.
Analysis Reports
================
The output files created by SHOWFREE make extensive use of PCBoard v14.5's
color @X codes. This means that a single file can be viewed by graphics
and non-graphics callers alike. Of course, the graphics callers will find
the file much more readable due to the advantage of color "cues" in reading
the graph.
Event Times
===========
Let me give an example in the way of explanation for the need for entering
your event time into the configuration file:
fact 1: system event starts at 2:00am
fact 2: system event ends at 2:30am
fact 3: all caller logs are deleted during the event
Given the above information you will be missing information in the caller logs
that would give the SHOWFREE program knowledge that the system was unavailable
from 2:00am until 2:30am. The system availability analysis would, therefore,
incorrectly show the time from 2:00-2:30 as being AVAILABLE when in fact it
is not.
If you run multiple events you most likely do NOT delete the callers logs in
each one. Therefore there is no need to list each event. Just the event that
deletes any trace of itself by deleting the caller logs needs to be shown.
Advanced Ideas
==============
One phone line can actually participate in more than one line group. For
example, the Salt Air BBS has only two really distinct groups. They are:
801-261-8976 HST modems - nodes 1, 2, 3, 7, 8, 9
801-261-8979 Hayes modems - nodes 5, 6
However, we have two phone lines that are private and cost more to access. We
call them our Extended Support lines. These two lines participate in the same
ring down used by the HST modems making the following true:
801-261-8977 Extended Support - nodes 1, 2, 3, 4, 7, 8, 9, 11
Because two additional nodes are in the group the system availability for this
particular group is higher than the system availability of the HST group.
That is because with more lines to dial into there is a greater chance that
one of them will be free and waiting for a phone call.
The Salt Air system, therefore, runs the SHOWFREE program three separate
times like this:
showfree hstnodes.cnf
showfree private.cnf
showfree node56.cnf
copy hstnodes.rpt+node56.rpt+private.rpt \pcb\gen\blt15
The last line, of course, copies the three separate reports into one single
bulletin called blt15 which is our bulletin #15 which shows system
availability charts for Salt Air.
Have fun!
=========
Well, the documentation sure ended up being long for such a simple program.
The output files created by SHOWFREE are fairly colorful and should provide
useful information to your callers letting them know when the best times are
to call your system. And once it is set up it runs itself forever so you
don't need to worry about it again. So sit back and relax and have some fun.
As is
=====
No waranties are given on this program. It is provided as is. It works on
Salt Air given the configuration we use. If your system is similar enough to
fall within the requirements mentioned in this documentation then the program
should work fine for you. There are no guarantees for bug fixes (if any are
needed) or updates to the software (in case of wishlist items).
History
=======
06/19/93 version 1.2 Bug Fixes
There was a bug in v1.1 which was modified to accomodate event
information from v15.0. The bug created a range of minutes marked
as "used" between the last caller and when the event ran.
Effectively, the system would appear extremely busy just prior to
the event running. This has been fixed.
Also, if the Monitor Missed Connections option was enabled, it was
not fully counting the time in between the last caller and when
a "missed connection" was detected.
05/06/93 version 1.1 Operational change
Modified the "minutes used" calculations so that instead of assuming
a "100% free minute" when the last logoff and next logon are 1
minute apart, it will now count that minute as not available at all.
Due to the use of $$LOGOFF.BAT files and other processing that must
occur AFTER the logoff but BEFORE the next login, it was more likely
to be less than even half of a minute between calls if the two calls
were only 1 minute apart. So it does not even attempt to average
it to 30 seconds or anything. It just counts the minute as used
and therefore unavailable.
Also, in v15.0 you can monitor missed connections. This can
provide additional information such as when the phone rang but a
connection was never establish, or when a connection was made but
the caller never logged in. By factoring this information into
the system usage, the system availability is closer to reality.
NOTE: see the v15.0 section up above for required PCBTEXT changes
and settings in PCBSetup.
01/11/92 version 1.0 Original Release