Modbus node performance testing

I did modbus performance testing by using modbus simulator and modbus read node.

I have read data every second and generated 3 days report and checked .
I could see some data loss so Is there any other node rather than modbus read I can use to read modbus data ?

Any help is appriciated thank you in advance.

with regards,
Amar

I'm afraid your question is far too vague. There are far too many variables to even begin to answer this. For example:

How do you know it was a problem with the modbus nodes in node-red and not the modbus device being buggy/unresponsive?

Which modbus nodes did you use?

Is this modbus serial or TCP?

Are you 100% certain the modbus device was online 100% of the time?

Are you 100% certain there was no other external network outage?

Are you 100% certain the computer running node-red did not go to sleep or restart?

Are you 100% certain your method of measuring performance is accurate and reliable?

Have you searched for modbus in the palette - there are other contribution nodes.


Lastly, what were the actual results of your performance measurement? How many packets did you drop?. In any case, you should make your flows fault tolerant as there are so many things can go wrong at any given moment.

I used node-red-contrib-modbus : modbus-read node and I am sure that it was 100% available all the time and the efficiency came around 95% for example Out of 100 , 5 packets were missing.

and I also searched for the other nodes in palette yes there are other nodes as well But i am not sure which node to use.

So i raised a question if is there any node which other people have used and they got a reading efficiency around 98-99%

You haven't stated how you measured performance or answered most of the other questions.

My money is on the problem being either in your performance measurement method or something downstream.

I am certain you know this but I will spell it out anyway, it is very possible and fairly common for a cheap hardware device to be buggy and occasionally unresponsive.

So how exactly did you measure performance? Did you use a network analyser?

So basically I generated data every second for 10 nodes and I calculated the time interval(seconds) between these three days and checked.

thats how I analysed the data ,
Is there any other ways to analyse?

Okay, now I suspect your use of nodes and how you have arranged your flow is the problem

My bet is you have these all in parallel.

Yes if you want true analysis, you would connect a network analyser and you would inspect the TCP packets sent Vs received. Then you could accurately lay blame.

Could you pls help me with the network analyser , I mean what kind of network analyser I should use.

Asuming Modbus TCP (you didnt state Serial/TCP), then wireshark or TCPDump are popular tools for network analysing.

You can save yourself a LOT of investigation and analysis if you tell us/show us how you are grabbing data. As I said above, I suspect you are doing multiple parallel reads based on the comment "10 nodes". This will create contention and inconsistent results & ultimately the cause of your problems.

You might want to read these 2 articles to help you design a better, faster, mor consistent flow for accessing data (skip straight to Part 2 for the juicy details)

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.