OPC Surrogate - Differences between System 800xA 5.1, 6.0 and 6.1 versions
Has the way OPC DA Clients access OPC Surrogate been modified in System 800xA 6.1 when compared to the old versions?
Some customers I have contact with are seeing (loss) performance differences when comparing versions 5.1 and 6.1, for example (customers who recently upgraded their Systems).
According the documentation, the limitations (in quantitative terms) remain the same...
Thanks for attention.
Voted best answer
Have you forgot to set the InhibitAudit registry key (that optionally, bypasses the need for performing pre-write-reads when AuditEvent_OperatorAction is enabled)?
Some ABB PLCs, e.g. 800xA for Advant Master controllers are sensitive for how the application is designed. E.g. there is a huge difference if the history logs collect on 9 seconds or 15 seconds. The latter is not making use of the cyclic data support in aforementioned brand of PLCs. You must stay on 1, 3 and 9 seconds update rate, or else the overall performance will drop, sharply. The same restriction applies to all subscribers, not just history logs.
We have (at least one) support case, still under investigation, were the root cause seem to be within the source OPC server (this time not 800xA for Advant Master) running below the 800xA System (i.e. the one associated with the brand of PLC used in the system). Logging have shown substantial delays in the server's time to expedite OPC actions after upgrading from v5.1.
There are more differences, e.g. 5.1 runs Win7/Server2008 while 6.1 run Win10/Server2016 that may come into play, e.g when using DCOM between different computers.
If you like, use afwapplogviewer to turn on the OPC Adapter Basic log to level 3 (the adapter runs inside the OPC DA Connector service interfacing the source OPC server) and compare what you see with the old vs the new version. There may be substantial amounts of logging output; I suggest starting in a small scale, e.g. with a test client adding just a single OPC group and item to learn the log format. Then connect the client experiencing "poor" performance. If possible, remove all other clients during the test or else you must find a method to filter for the items being handled "slow".
I have attached a document where I have made some examples of use cases with applog.
With some effort, you can use the logging timestamps to calculate the time it takes to, e.g. add items or perform write operations. Personally, I use a blend of shellscript (e.g. awk, Notepad++ and Excel).
If you provide me with the logging results I will try to reply with comments (I'm sure you can find me in the ABB email directory). Please make proper annotations when logging; don't just send a 1GB log file with no comments...
Please note that applog is an R&D tool not intended for use by end customer; there is no documentation you can fall back on, known bugs, etc. I therefore recommend that you file a support case with your regional ABB support center to obtain professional assistance with your problem.