Hi We have a scenario where we have to Post a comment json at every step in our work flow. calls count will be very high,When I generate a Health check which is in High risk which is taking 25.2% of all CPU time, 22.9% of all operations, 8332 mins of CPU time, 108948293 operation count.
we are not doing any other operations in that PM just call integration.
Any suggestions to minimize this CPU time or how to improve performance?
Discussion posts and replies are publicly visible
That's a lot. Did you consider to store these comments to DB and then do async batch calls to send them over?
Thanks for the quick response,
Actually we are storing the DB and sending to external system. what ever the actions or steps happening in Appian we should inform our external system. with this post call. so we can not do batch process here .
Well, in the end, you have to do what you have to do. The health check is about risk management. If you cannot change the design of this, you need to accept the risk, making sure it does not turn into a real problem.
But I really want to question whether you need a real time notification to that other system.
Yes Stefan Helzle you are correct.
business need real time notification at every step.so we are taking this risk. which is causing our process slowness and Performance issues ..
if any suggestions always welcome.
thanks
Swaraj Kumar
What is the service at the end of the integration doing? Is it processing your message in real-time and forcing you to wait until the message is processed completely, or can it accept/store your message unprocessed and process it offline? The first scenario will keep your node active until the processing is complete or it times out which increases the burden on your server.
Hello Stewart Burchell,,
Not sure about their system side but our side we are using Post method to post our Json.
The Post is synchronous but if they're holding the connection open until they've finished processing then that might be an opportunity to optimise. If the take you message and store it, close the connection and then process the message asynchronously that would be better.