At ING, we run four instances of RA, each on their own linux infra. These instances can be rather busy, and we experience delays in releases and timeout errors in the webserver.
We did increase the default heap size in the catalina.sh for the Automation Center from 2GB to 4GB. However, we wonder whether any of you out there have any experience with tuning RA further, changing settings, or adding CPU, and so on, and how successful that was in improving performance?
And secondly, any ideas about how many (concurrent) releases are feasible? Is there a practical limit and where do you guess is lies?