Hello,
How many devices of Modbus TCP drivers can be created?
Can I have 77 modbus TCP devices(I will read around 260 tags from each PLC (20.000 OPC tags in total))?
In the project I have around 160.000 tags( Memory tags, expression tags, OPC tags…)
I am using Ignition 7.9.7 (64 bit)
Thank you very much!
Yes, no problem.
There's no intentional limit to the number of devices, but the current design probably makes the practical limit somewhere around a couple thousand.
Hi!,
When you say “a couple thousand” did you refers to “devices” or “OPC tags”?
Thank you!
I'm talking about a couple thousand devices.
Hi,
So It will possible to have 20.000 OPC tags (distributed in 77 devices)?
Thanks,
Yes, that shouldn't be a problem.
Ok, thank you so much!
Hi Kevin,
I have 3500 Modbus RTU-OverTCP Devices with 250 TAGs each at 30sec scan rate with 20% of the TAGs changing in every scan.
I'm thinking of an architecture of 5 backend servers and 5 frontend servers, what do you think?
Thanks
That should be plenty. If the backend servers are strictly just acting as OPC servers or IO servers or whatever then you might not even need 5.
How many frontend servers you need is not really related to the number of devices.
What actually worries me is that it's RTU-over-TCP - this tends to be a pretty unreliable setup...
Thanks for the answer Kevin.
Let me see if I understand correctly. Are you saying that an Ignition Gateway acting as an IO server (platform+tagHistorian+alarm) is capable of collecting data from at least 1000 MbusRTU devices without any issues?
Yes, the only real issue with 1000 devices would be how difficult it is to configure and manage them, and perhaps how much log entries they will generate.
But it's not uncommon that we see Gateways with ~1000 Modbus devices.
hummm what do you mean with "unreliable"?
Modbus RTU over TCP is basically a made up variation of Modbus that takes an inherently unframed payload (Modbus RTU) and puts it on a transport that requires framing.
So implementations have to be heuristic / state machine driven.
On top of this, it usually means you're talking over something like a brainless ethernet-to-serial unit or modem.
and there's absolutely no protection from accidentally polling from multiple masters, which is absolutely not supported with RTU-over-TCP.
To make Modbus RTU over TCP even close to reliable, you need to configure the ethernet to serial converter to buffer properly, or you simply cannot meet RTU timing standards. See the appendix for Moxa devices in my alternate driver's user manual.
(My alternate driver is also much nicer to your gateway logs.)
Okay, I understand your point, and you're 100% right.
What if I use Modbus Bridge to convert from ModbusRTU to ModbusTCP instead of an Ethernet-to-serial device?
That's much more reliable, but at 3500 devices may be cost prohibitive unless you have them already...
Luckily, we already have them. We use RDL3000 Redline Radios, which are capable of functioning as Ethernet-to-serial devices or Modbus bridges.
Hi Phil, you mean the Advanced Modbus driver?
Yes.