Installation guide
64 Estimating Network Performance
890 USE 100 00
Two additional operations are occurring on the network. Computer B is
in a programming or load/record/verify operation with controller 3.
Plant personnel accessing any node from the P230 programming panels
using the controllers’ built-in Modbus to Modbus Plus bridge mode.
These operations are handling data that is not currently used in the
application. Their activity on the network will, however, affect the
response times for the first two operations. For example, the network’s
token must be held for a period of time by Computer B and controller 3
while they transact the necessary data. If this operation terminates,
and these nodes have no other immediate operations, each will retain
the token for only a minimum time before passing it. (Methods for
estimating token rotation time are presented later in this chapter.)
3.4.2 Planning Your Application Program
The way in which you program your application will also affect network
performance. For example, the approximate time required between two
controllers to request data and receive it is:
V One token rotation time for network access to send the request
V One scan time in the receiving controller to process the request
V One token rotation time for network access to send the response
V Two scan times in the initiating controller to process the response
If data were transferred as a global transaction, it is received by
multiple nodes during a single token pass. The approximate time for
this is:
V One token rotation time for network access to send the global data
V One scan time in each receiving controller to process the global data
Your choice of polling or unsolicited transactions will greatly affect
network performance. If you construct your application program using
polling techniques, you will force the network to handle some quantity
of transactions that do not return data. This will tend to increase the
aggregate amount of network traffic, and will diminish the ability of
the network devices to manage their data paths and to acquire the
token.
Rather than using polling techniques, you can gain improved message
throughput by implementing eventdriven read or write operations
between the devices. Receiving devices can be prepared for unsolicited
data by having their applications sample ‘flag’ bits (bits that are