tvey
Lifetime Supporting Member
Is anyone aware of any performance limitations with Wonderware's DASSIDirect IO server?
Here's the scenario:
- Client has a WW9.0 app with approximately 41000 tags. PLC is an S7-400 series.
- There is ONE IO server running SMC Console w/ DASSIDirect. (There is a redundant failover IO server as well, but only 1 is active at once).
- DASSIDirect is configured with two device groups: S7PLC (1000ms update interval) and S7PLC_Slow (2000ms update interval).
- There are 10 client View nodes that all point to the same DASSIDirect service on the main IO server. Tags are fairly evenly split between two Access Names (S7PLC and S7PLC_Slow). Currently both access names are set to 'Advise All' instead of 'Advise Only Active Items'.
- There are intermittent communication problems with this system. The communications between the IO server and the PLC occasionally time out (IOStatus=0).
I am wondering if anybody is aware of any 'hard' limitations with the DASSIDirect service.
For example, could the use of 41000 tags in 2 device groups be a problem? Could there be a performance gain if we were to split the tags across more device groups with staggered poll intervals?
How about on the Wonderware side? Would there be any performance gain if we split tags across more access names (still ultimately pointing back to the same DASSIDirect service?).
How about pointing 10 clients at the single IO server - what does it take to stress DASSIDirect?
It's fairly straight-forward to test out any or all of these changes, but I'd like to make sure we're not running up against some fundamental issue here to start with.
Thanks,
Trevor
Here's the scenario:
- Client has a WW9.0 app with approximately 41000 tags. PLC is an S7-400 series.
- There is ONE IO server running SMC Console w/ DASSIDirect. (There is a redundant failover IO server as well, but only 1 is active at once).
- DASSIDirect is configured with two device groups: S7PLC (1000ms update interval) and S7PLC_Slow (2000ms update interval).
- There are 10 client View nodes that all point to the same DASSIDirect service on the main IO server. Tags are fairly evenly split between two Access Names (S7PLC and S7PLC_Slow). Currently both access names are set to 'Advise All' instead of 'Advise Only Active Items'.
- There are intermittent communication problems with this system. The communications between the IO server and the PLC occasionally time out (IOStatus=0).
I am wondering if anybody is aware of any 'hard' limitations with the DASSIDirect service.
For example, could the use of 41000 tags in 2 device groups be a problem? Could there be a performance gain if we were to split the tags across more device groups with staggered poll intervals?
How about on the Wonderware side? Would there be any performance gain if we split tags across more access names (still ultimately pointing back to the same DASSIDirect service?).
How about pointing 10 clients at the single IO server - what does it take to stress DASSIDirect?
It's fairly straight-forward to test out any or all of these changes, but I'd like to make sure we're not running up against some fundamental issue here to start with.
Thanks,
Trevor