home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Cream of the Crop 1
/
Cream of the Crop 1.iso
/
EDUCATE
/
TN210.ARJ
/
NETWORK.EXE
/
NET-6.TXT
< prev
next >
Wrap
Text File
|
1992-07-02
|
5KB
|
81 lines
NET-6.TXT
OPERATING PRACTICES
-------------------
No matter how efficient a network configuration may be, poor operating
practices can create congestion. Since simplex networks perform best when
lightly loaded, more "mileage" can be gained from them if the users were to
employ common sense in this regard and practice PACKET CONSERVATION.
Just as world-wide attention and awareness is being focused on the fact that
our natural resources are dwindling, we packeteers could benefit if attention
were focused on the impact our actions make on the network. Since the network
is limited, it would be well to keep in mind every connect sent through the
network could adversely impact someone else either up or down the line. This
isn't to say one shouldn't use and enjoy the network. But such use should be
adjusted in terms of consideration to others.
With the price of IBM XT clones under the $350.00 mark, it's easy for someone
to install a packet BBS these days. Or they can load it with TCP/IP
software and add a node to the network. Unfortunately this is quite frequently
done without giving any thought as to what impact it will make. It may well be
the addition of another BBS in an area that already has adequate BBS service,
could overload the network and thus cause dissention among the users.
Frequently the operation of a TCP/IP node is harmful to the network function.
This is true even when the TCP/IP node is not in active session. Remember the
node barf spoken of previously? TCP/IP nodes add to the "node chatter" found
in a dynamic network. This added overhead steals away circuit time available
for other users. Network nodes provide connectivity and routes that benefit
all users. Usually not true with TCP/IP nodes. The majority of TCP/IP nodes
are nothing more than an end terminal for its owner. When in active session,
TCP/IP nodes can adversely impact the "hidden transmitter syndrome". In the
hands of an inexperienced person, operation of TCP/IP through the system can
disconnect all users. An active node can prevent other usage for hours on end.
We are not suggesting TCP/IP is bad. TCP/IP has many features that works well
when used in its intended environment. This envirnoment is on a robust
full-duplex 9600 baud or higher, circuit. Originally intended to overcome
varying protocol compatibility problems on LL systems, TCP/IP has higher over-
head than AX.25. As such it will never perform as efficiently. Using it on a
multi-user simplex 1200 baud radio channel frequently causes problems to other
users.
Continuing with the concept that we users can modify our operating practices to
make the present network more efficient, lets review and expand on the points
made previously.
We know simplex networks perform best under light to moderately loaded
conditions. If we all were to practice "packet conservation", it stands to
reason we will be able to get more efficient use out of our present systems.
By packet conservation, I am referring primarily to avoiding the little verbose
habits we tend to have. One of these (which is covered in nearly EVERY
packet primer) is beaconing. There are exceptions to this, but in general it's
better to NOT beacon. If one really feels a specific need to beacon, than it
should be limited to intervals of not less than once every 15 minutes. Along
simplex networks, user beacon transmissions may cause collisions with network
nodes, thus QRMing the channel and delaying throughput. On a full-duplex
repeater, beacons will delay ongoing traffic through the repeater.
An extreme case of beaconing was noted a few years ago off one of the local
nodes. Several BBSes had converged on the channel and apparently the SysOps
felt the need to compete with each other. The result was various long beacons.
Some of cartoon characters, some of page-length announcements, and each seemed
to have a repetition rate of every two or three minutes. Inner-mingling the
beacons was near constant BBS fwding activity. After a time, the few remaining
users were oblivious to the beacon activity since they filtered the offending
BBS callsigns out. The net result was the beacons and verbose connect
messages were counter-productive inasmuch as many BBS users were driven away
due to the congestion. Those who remained weren't able to enjoy ragchew
activity and went into non-monitor status.
In this case, too many BBSes sending too many beacons on the same frequency
lead to the decline of local packet activity. It would then seem prudent
that potential BBS SYSOPS make a determination whether additional BBS services
are required on channels already occupied by BBSes. Factors influencing this
determination would include network policy if any, potential user base, lack
of specialized BBS services, and adequacy of BBS forwarding channels. With
the availability of multi-connect BBS software, one full-time dedicated BBS
in an area per frequency should be sufficient .