How many LMs I need for such a large project?

[gdimaria]

Hi,
I have to manage a large building whose ETS project consists of about 8000 objects, including rgb and Dali lights, shutters, thermostats and presence detectors. Also, I have to run many scenes (at least 60).

My idea is to use 4 LM, one for each section of the building, and a fifth LM as master supervisor.

I would like to know if my idea is valid and how to best divide the load: I have to load on each single machine only the objects (and scenarios) that will manage or the whole project?

In short, I would be grateful if you could suggest me the best solution to manage everything without problems.

Thank you.


Peppe

[admin]

We do not recommend more than 2000 objects per LM. Your setup should work if master LM does not require to get all values from all objects. You will need to correctly configure KNX/IP filtering table to prevent overloading both LM and KNX/TP lines. From reliability point of view it's better to run all local stuff like scenes on each line separately and use one master LM for visualization.

[Tokatubs]

@Admin

I am kinda in the same situation. As you can see i am way over the limit that you write of 2000 objects. I have not seen any trouble but guess investing in one or two more is needed. 

Is there any example or something like that to see how to setup multiple LM together. ?

   

[admin]

I'm not saying that 2000 is a hard limit. There's no universal solution here as there are many different scenarios. For example, you can have high CPU load even with a small number of objects but with high KNX bus load and many scripts. For bigger projects it's more convenient and reliable to split the load between several LMs.

[gdimaria]

We do not recommend more than 2000 objects per LM. Your setup should work if master LM does not require to get all values from all objects. You will need to correctly configure KNX/IP filtering table to prevent overloading both LM and KNX/TP lines. From reliability point of view it's better to run all local stuff like scenes on each line separately and use one master LM for visualization.

So, if I understand, each LM must have only its vis. structure and the related objects it uses to run its own scenes and/or scripts, right?

The master LM doesn't need objects at all, but just visualize others in frames, right?

thanks

Peppe

[Daniel]

I think admin meant that objects which are needed for visu goes to Master LM but I know about projects build in the way you explained. I would make admin way if reasonable amount of objects are needed to make full visu and your way if too many. You could also make a combo of both.

[gdimaria]

I think admin meant that objects which are needed for visu goes to Master LM but I know about projects build in the way you explained.  I would make admin way if reasonable amount of objects are needed to make full visu and your way if too many. You could also make a combo of both.

Sorry, Daniel. I don't think so, because for Visu on the master you just need 4 frames with the IP address of the other machines. So I can't understand your idea of a "combo"... maybe it's better to wait for Admin reply.
Anyway thanks for your reply.  

[admin]

What Daniel said is correct. All LMs devices exchange data via KNX/IP with filtering tables applied in a way that only required data is passed. This also applies to the master/visualization-only LM because I doubt that you need all 8000 objects there.

[gdimaria]

What Daniel said is correct. All LMs devices exchange data via KNX/IP with filtering tables applied in a way that only required data is passed. This also applies to the master/visualization-only LM because I doubt that you need all 8000 objects there.

Well, the point is that most of the objects will be involved in the visualization. That's why, referring to Schneider's "Addressable limits of the Wiser for KNX controller" guide, (https://www.se.com/it/it/download/document/AN017/), giving it as valid for LM too, I had thought of using the master only as a viewer of a couple of pages with only frames in which to display the other 4 LMs, each one running its own scenarios and scripts.

Summarize my idea:

- 4 LM each with its objects, visu, scripts, scenes.
- 1 Master LM, without objects and only with a few visu pages with 4 frames to recall the other LMs.

What I do you think?

Thanks
Peppe

[Daniel]

As I pointed earlier, I know projects which were done this way. On other hand you could try to build full visu in master LM and if load will be too big you can always export parts of visu to each individual LM. In Vis structure you have export possibility.

[gdimaria]

sorry, the schneider guide I referred to is not the one mentioned above but "AN042_Optimization of visualization in complex projects" ( I can't find the link, neither attach it)

Thank you.

Peppe

[Daniel]

I was helping to write it

[gdimaria]

I was helping to write it

ok, found it!

https://download.schneider-electric.com/...f=AN042_SL

bye