We have an ODH historian with around 6000 tags and which is requesting data from the App-node (TPN server which is on the LCN network). We are using OPC Optimizer to help with server load. The problem is that the LCN is being overloaded and is very slow.

On OPC Optimizer we have configured 500 items per group. A group is the update rate of the tags-so if there are 1000, 1 sec tags ...it is one group but since we only have 500 per group ...the optimizer would split them into two groups. The number of groups to connect with the tpn is set to 1. Which means instead of having more than one group request data with one connection, each group is initiating its own connection with the tpn sever. Basically around 13 connections are being made to the tpn server from optimizer with 500 tags each. I think there are too many requests to the app-node which is overloading the server which is somehow overloading the lcn.

I have a few questions. Will more connections from the optimizer to the tpn server slow down the network (LCN)? Or could it be that requesting data for 6000 tags is too much bandwidth on the lcn? Any suggestions are very appreciated.

Also if a server is working hard, does that slwo down the network?

Thanks alot

 

We are fetching 15000 parameters per minute with our PHD system. We have 2 history modules, 3 application modules and 20 stations on that system and find it not to be slow.

You mention the LCN being overloaded. Did you actually do the LCN load calculations?

Did you monitor the processor status data point parameters to figure out if your problem might not be caused by a certain NIM (max 1500pps), UCN, HPMM or even I/O link?

What else is going on on your LCN ? Are you also historizing using a history module?

Do you have application module applications running fetching a lot of data and could these be slowed down?

How many displays are open at any given time (we once found an engineering station with 10 displays open looking at the same HPM)? and are all parameters being fetched regularly? For NAME, PTDESC, EUDESC etc we configured the displays to only fetch on call-up.

For the TPN server it self. Is your communication optimized so you send a request per data owner? If it isn't each data owner needs to respond to multiple requests - overhead included.

Is the TPN server configured optimally for the type of communication you want to do? (typically lots of small requests as opposed to not so many big requests)

Maybe you can have Honeywell do a system performance baseline test which helps you to find bottlenecks.

 

I am very interested to know about your OPC Architecture. OPC Optimizer should help you reduce the load on the server. Depending on how you are using it, there are many features in optimizer that can do the trick for you. I recommend you call support immediately to assist you.

+1-877-OPC-4-ALL
+1-780-231-9480
+1-780-264-6714

Wassim Daoud
Global Solutions Architect
MatrikonOPC

 

I have a rticket with matrikon. after reading more about tpn and optimizer, i found that the number of items per group and the number of groups per connection would affect the lcn network but i am not sure how. i found this information at eric murphys website:
www.blogs.matrikonopc.com

 

>I am very interested to know about your OPC Architecture. OPC Optimizer should help you reduce the load on the server. Depending on how you are using it, <

Also i looked at the log an dit showed deadlock connections ...

 

Hi Lalas,

My name is Gilbert Waniandy and I am the Support Manager for MatrikonOPC.

I would like to resolve your issue for you but have not had any luck finding your ticket in my system under your forum name.

If you could, please contact me directly at +1-780-233-0196 and I will do my best to answer all of your questions.

Regards,
Gilbert Waniandy
MatrikonOPC Global Support Manager