We were recently tasked with helping an application support team mitigate a problem while they troubleshoot the code. This particular application is on 5 web servers all load balanced. When the application pool reaches 1 gb, the application starts to struggle.
We were asked to monitor the app poll and then when the threshold is reached to rename the heart beat page, thus pulling the application out of rotation and allowing the app pool to release resources. Once the resources release to around 100 mb, we rename the heartbeat page again and thus back into load balancing.
After experimenting with Orchestrator and SCOJobRunner, we were able to accomplish this. We have basically come to the conclusion that this combination will allow us to automate almost everything in our environment.