I'm currently in the process of translating an in-house developed SCADA solution to Ignition. In all previous projects using Allen-Bradley PLC's I've just used Ignition's built-in Logix Driver to connect to the PLC without any discernable issues. For this project our group has been told by a Rockwell rep that the PLC in question has a limit to the number of tags that can be referenced using this connection method and they recommended using FTLinx Gateway instead. I've never heard of this limitation before. Is there any validity to this?
Yes, especially if you'll be accessing AOIs and pre-defined types, or your application is otherwise not optimally structured for efficient access by the drivers.
It's not a hard limit on the number of tags, it's more that you'll max out the comms capability of the PLC sooner than later in these circumstances.
FTLinx has some proprietary access methods that perform better, including some that even Phil's driver doesn't support. (his driver does have reverse engineered support for reading AOIs efficiently as long as they contain no members that aren't externally accessible)
It is also possible that the Rockwell rep is confused, and is citing the limits in the PLC's OPC UA support.
I'd bet this is likely the case and as soon OPC-UA is brought up, Rockwell thinks of the only way to do OPC-UA with a ControlLogix is in v36+ and that's very limited. You always have to keep in mind with these vendors that they know their products well, but usually know almost nothing about anything else.
I know this thread is from last year, but I figured I'd add this in case someone else comes across it while researching the same issue.
I think the distinction between the Logix driver limits and the controller's native OPC UA limits is probably the important part here.
The built-in Ignition Logix driver isn't using the PLC's native OPC UA server, so the OPC UA node limits on a newer ControlLogix aren't automatically a limit on how many tags Ignition can reference through the Logix driver. At some point it becomes more about communication load and how efficiently the data is being accessed than a simple fixed tag count.
I was trying to understand those ControlLogix OPC UA limits recently and came across this:
What helped me was the section on the Logix side specifically. Native OPC UA on the 5580/5380 starts with v36, the controllers have actual node limits depending on the processor, and OPC UA shares the controller's communication resources with CIP Class 3 traffic. So I could definitely see how a conversation about those limits could get generalized into "there's a tag limit," even though that's not really the same thing as using Ignition's native Logix driver.
FTLinx may still have an advantage for some applications because of Rockwell's proprietary access methods, especially with AOIs/UDTs as mentioned above, but I'd want to know the exact controller model and what limit the Rockwell rep was referring to before changing the architecture just because of a claimed tag-count ceiling.
Hopefully this helps someone else who finds this thread later and is trying to make sense of the same distinction.
There are tradeoffs with either path.
Linx GW is fast and your tag paths don't change when you update firmware. So if you use Linx GW to communicate with pre-20 PLCs you will not have to update your device tagpaths in Ignition after an upgrade.
One big downside for Linx GW is that it's not cross platform and I have seen windows updates break it. I had a situation where a plant's Linx GW stopped communicating until a reboot. I did a Rockwell patch rollup and it hasn't done that since but it is a notable downside. I've also noticed that it generally doesn't see new tags with a tag browse till after a reboot. You can still manually enter them and they will work but the tag browse will not see them.
Key take away. The thing that makes Linx GW fast is also what makes it unstable.
The biggest upsides of the Ignition OPC UA driver is that it's cross platform and stable.
The only real downside is the official public spec that it is (and must be) built from is slower than the secret sauce Rockwell does with Linx GW. Well that and the Class 3 traffic as you already pointed out.
I feel like there's a time and place for either solution.