Specifications
Allen-Bradley ControlLogix Ethernet Driver Help
3. Server tags referencing a Structure Logix tag composed of one-third or more of the Structure tag should
be assigned to Logical Blocking devices.
4. If Symbolic Mode is used, Logix names should be kept to a minimum length.
5. Logix arrays should be used as often as possible.
6. Only the necessary amount of System Overhead Time Slice for Ladder Logic/FBD should allocated to leave
the rest for driver communications.
7. For projects that read a large number of string tags in Logical Mode, uncheck the "Automatically Read
String Length" option located in the Logix Options tab of Device Properties.
For information on increasing driver statistic results, refer to the instructions below. For more information, refer
to Optimizing Your Application.
1. Devices should be spread across channels. More than one device should not be put on a channel unless
necessary.
2. Load should be spread evenly across devices. A single device should not be overloaded unless necessary.
3. The same Logix tag should not be referenced across different devices.
Note: Although these general rules can help optimize performance, it ultimately depends on the application. The
scan rate can obscure results: if tag requests are light, read and write transactions can complete before the next
request comes in. In this case, Logical Blocking and Logical Non-Blocking will have the same Performance
Statistics results. If tag requests are high (many tags or high scan rates), transaction completion time may take
longer. This is when the strengths and weaknesses of Logical Blocking and Logical Non-Blocking become
apparent. Performance Statistics can help tune the application for maximum performance. For an example, refer
to Performance Tuning Example.
Performance Tuning Example
Statistics can be applied to any application. In the example below, the Quick Client is used in the performance
tuning process. The idea is that all the tags used in the project are read at the same time at a fast scan rate.
Although this is not realistic, it does provide an accurate benchmark to the project layout in the server (tags
belonging to specific devices, devices belonging to specific channels, and so forth).
The statistics gathered are relative. Users should start with a server project layout, gather the statistics, and
then tune. It is recommended that more than one trial be used to properly assess the results for a given layout.
Once the most efficient layout is determined, the client application can be built with reassurance that the server
is optimal.
Caution: Performance results obtained using the Quick Client do not equate to performance results obtained
using a client application:several factors produce discrepancies. Although performance tuning with the client
application is more accurate than with the Quick Client, the tuning required not only affects the server project,
but the client application as well. It is recommended that the Quick Client be used to tune the application before
the client application is developed.
Note: The tuning process described below assumes that all tags are being read at a fast scan rate. Writes hinder
the performance.
1. In the controller project displayed below, there are the following:
2 Atomics
1 Atomic Array
1 UDT
1 UDT Array
1 Pre-Defined Type
Note: Overhead Time Slice (OTS) = 10%.
www. kepware.com
35