About Mango capacity limit
-
I want to know whether mango can meet my project capacity request. I try free version, and connect to 1 PLC by modbus ip, to poll 20 points, and I set poll interval is 500ms, and log to database every 3 minutes, it running well, but CPU keep busy, most of time 80%~100% occupied, and harddisk also busy, keep lighting flash, bandwidth arround 200KB/S, my computer is i5, 8G RAM. so I worry about if mango is possible to handle 50 PLC (1000 points, each poll interval 500ms) by same time? does mango has ever tested about maximum capacity limit? any proposals for this? thanks
-
by the way, I can't make modbus persist tcp to work, it just poll data one time when system restart, and then do not update anymore. so currently i use modbus tcp connection, it work well for one PLC.
-
You might be interested in this https://infiniteautomation.com/large-scale-mango-installation/
You should be able to run 50,000 data points for testing on your computer so there is no reason why 20 points logging every 3 minutes.
You might also check the data source performance: https://help.infiniteautomation.com/data-source-performance/
Also, make sure you are using Oracle JDK.
You can also look at the System Status page in Mango to see how busy it is and view all the internal performance metrics:
https://help.infiniteautomation.com/internal-task-performance/ -
@timkk880 said in About Mango capacity limit:
modbus persist tcp
Did you mean Mango Persistent TCP? That's the data source side of the publisher with the same name, which is for sending and synchronizing data between Mango installations.
-
thank you, JoelHaggar, i will look into it.
sorry, phildunlap, I mean Modbus IP TCP keep alive. it do not work. anyway, Modbus IP TCP work well.
I have a question, what's the speed limit for modbust IP poll interval? 1ms or 50ms or 500ms or 1 second? -
That depends on the device you are polling and the device you are polling from. Mango offers options down to 1 millisecond but we don't really recommend trying to poll that quickly, and instead say you should be timestamping your values at your sensor device if you need that precision. You can always try it out, though!
-
thanks, phildunlap