Monitoring Node Communication CPU Load

INFI 90 systems are characterized by PCUs that are Nodes on a Ring.  Traditional systems include a NIS/NPM pair that handles the communication between the Modules on the PCU/Node and the rest of the system.  The performance of this interface can be monitored trivially using WatchWindow in DBDOC Hyperview.

Note that ABB Technical Bulletin TB1999054A gives details on the various performance statistics available from a node.  It also specifically mentions that this capability was introduced in INNPM01 firmware revision C.1, so it is not available in older versions.  It being at least a decade and a half since that firmware revision was introduced, let's hope nobody will be disappointed by not being able to do this.

Using DBDOC Watch Window to Monitor Communication CPU Usage

To see what percent of the communication module's processing power is being used, you simply monitor Loop L, PCU P, Module 0, Block 11.  The example shows Loop 1 PCU 4 being monitored and the monitoring of Loop 1 PCU 6 being defined.

Notes about this monitoring:

1. This fetch is never done in turbo mode, so you only have 10 to 15 values per second available, and these are shared with all your other Hyperview users.  
2. Running for one or two "Tmax" periods for exception report flushing should give you a good picture.
3. Turn off the Watch Window data collection by clicking on the green clock icon when you have a picture of the loading.
4. If you do a save of a module in the PCU while you are monitoring, you will get an idea of the maximum communication loading possible in the node.

Interpretation of the Data

Experimentation has shown that CPU loading above 90% shows a situation where exception reports (XRs) can be delayed or lost.  This is the bad side.

What about if you find the loading very low - under 10%? This is the good side. We believe this means you have a lot more exception report generation capability than you are using. Your system is capable of significantly better performance with respect to tighter values for the HMI and for the history system. Pushing the communication usage out of the idling range will potentially be very beneficial.

You might refer to Improved Tag and History Data Precision in INFI 90 Systems for more on what you can do to understand what you might be suffering from and how to get more out of your system.

Results

We would be happy to get a copy of the Watch Window data, which is in a .CSV file that is created by the monitoring.

Disclaimer

The monitoring indicated has been done on dozens of DBDOC systems with no known problems.  It is the same monitoring as done by Process Portal B and gets the same result (of course).  Monitoring only one block makes the load as minimal as possible.  However, if you have any question or hesitation, your consultation with your ABB support specialist should give you more insight.  We are very willing to work with your experts.