Tridium Bacnet polling Speed Tracer

[BMTXdave]

Has any one had any problems with tridium bac net polling, I have a site with 15 BCU's were we use a tridium interface for a web server, after investagation with wire shark protocol anlyser found extensive bacnet traffic from tridium too devices (Approx 1000 scc vav controllers spread accros BCU's ) finding after 2 weeks of operation BCU's lock up NOTE: BCU Processor Idle Time @ around 1500 % and VM 35% NVM 40%, and when you go to access VAV from tracer,the database go's out of sync and tracer loses comms with the scc and freezes for approx 5 minutes and then tracer recovers but comms is extremly slow in mini monitor "Regersting comm 5 work station for installation" appears. After this if i clear ram reset BCU, all scc devices and BCU return to normal Processor idle time 65% works fine for another 2 weeks, i beleive this is from excessive polling from tridium, has any one come accross this or are there any new service packs out there for tracer/Tridium? I am running latest image 17.33 on version 17 sp2

[willf650]

You don't have the latest BCU image but it is doubtful that it will fix anything. 17.36 is the current image and has been around for about 6 months. I've never personally witnessed it as my Tridium expierence is limited but I have heard from a different manufacturer that the polling speed of a jace was messing up thier equipment as well on the BACnet side. I know this is obvious but have you tried turning down the poll frequency of the jace. I now have a few sites with a BCU serving up points to a jace but not the size of yours and haven't had issues thus far. Biggest site only has 8 BCU but the Jace is not used as a webserver on this site. If you come up with an answer let me in on it because I would be interested to hear the answer.

[ntec1]

Try using pollOnDemand containers also.

[BMTXdave]

Sorry i have not had a great deal of experience with tridium, is polling on demand the same as change in state polling ? and were could i find this parameter, i know base level system archecture and and point configuration
stuff just not high level programing.

P.S thanks everyone for the fast response this problem is a pain

Also i am from trane australia we are a little slow in getting latest updates

[BMTXdave]

Hey willf650 could you email me the latest service pack or just the image if possiable
ronburgunday@hotmail.com
I work for trane australia and we are a bit behind the times here

[BAS_Guy]

I would change everything important to poll about every 10 ~ 15min and everything critical to change of state. (COS) I would assume you should only have about 3 ~ 5 items per device that are COS.
I would reset the BCU idle time, VM & NVM on the BCU after doing the above and watch the idle time for a couple of weeks.
I would also remove any unneeded trends & alarms from the BCU's.

Good Luck

[BMTXdave]

There are approx 40 vav per floor 3 floors per BCU so each BCU is has approx 120 points all these points are polled on chang of state why only poll 3-5 on change of state

[checkvalve]

I'm am not very familiar with the setup screens for the BCU, but I know of a recent site where they were having a lot of trouble like you are describing. Apparently there is a check box in one of the setup screens for the BCUs to allow who has responses. Whatever the default setting is, it was causing the BCUs to overload each other.

The polling setup varies depending on whether you are using the R2 or Ax version.

If you are using R2 then most drivers constantly poll their points. Using poll on demand containers can greatly reduce the polling load on the network. You should also adjust the poll cycle time for the network.

If you are using Ax then all drivers are poll on demand. The default poll cycle time is 5 seconds, so if it hasn't been adjusted then you probably just need to adjust it to a reasonable time like 30 seconds.

If the BCU is setup to use COV/COS then assigning points in the jace to use that could reduce the traffic as well. Also make sure that the BCU and jace are setup to use read property multiple, which can significantly reduce the traffic as well.

[BAS_Guy]

There are approx 40 vav per floor 3 floors per BCU so each BCU is has approx 120 points all these points are polled on chang of state why only poll 3-5 on change of state

Are you only taking 1 point per VAV? If so you must have no room in the BCU because this is not alot of data. Thinking about it I am not sure a BCU supports COS/COV. I would reset the idle time and see what happens to it and how fast it changes.