home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Tech Arsenal 1
/
Tech Arsenal (Arsenal Computer).ISO
/
tek-20
/
tn210.zip
/
NETWORK.EXE
/
NET-3.TXT
< prev
next >
Wrap
Text File
|
1992-07-02
|
3KB
|
44 lines
NET-3.TXT
NON-USER ACCESSED BACKBONES
---------------------------
Thusfar several simple things has been mentioned that could improve throughput
on existing networks. These were:
1. Restricting users from accessing backbone trunks.
2. Decreasing the node radio TXD time on backbone trunks.
3. Incourage LAN users to upgrade to radios with shorter TXD requirements.
4. Installing DCD modifications to node AND user TNCs.
No doubt the hackles of more than a few are starting to rise over the idea of
being denied direct access to the backbone trunk, so let's talk about it.
First, let's keep in mind here the overall goal is to have a network that's
fast and efficient. The ideal network should be reliable and give little
delay from the time a user accesses it until all traffic is passed.
If node-to-node TXDs are optimized for minimum time-delay, on-channel users
with less than optimum TXD will create node-to-user retries, thus slowing
internode traffic. Additionally, the user's transmitter is apt to create a
"hidden transmitter syndrome" which causes collisions and subsequent retries
between the trunking node and possibly to other nodes on a simplex backbone.
This latter situation can occur even if the user's TXD does match that of the
uplinking backbone node. In terms of networking efficiency then, there are
several valid reasons why throughput is higher on a network backbone trunk
which denies direct user access.
Once attention has been paid to backbone trunk efficiency, how does one avoid
creating bottlenecks on busy LAN user access nodes? The solution on some
systems has been to install full duplex packet repeaters. They usually employ
the standard 600 KHz frequency split. Repeater receivers and transmitters are
not dependent on each other so they can react much faster than a simplex node.
The collision/retry problem associated with the "hidden transmitter syndrome"
is greatly alleviated since the repeater acts like a traffic cop. The result
is substantially higher throughput.
Where an access node is located in a high-usage area, a full-duplex packet
repeater/node is an efficient solution. The drawback of a packet repeater is
cost, since either a duplexer or dual antenna system is required. However at
shared sites with high levels of RF congestion, filtering may be required
anyway. In lower-usage areas, a standard simplex access node is satisfactory.