We have integrated the BTG extension, and added some personalization rules. We had a high DB load problem and disabled the rules. [ similar to https://answers.sap.com/questions/12750215/performance-issue-with-btg-evaluation.html ]
However, now we are facing other performance problems... Sometimes, some certain nodes from the cluster have a high CPU usage. We identified that the threads using much CPU
PID LWP CPU
17668 24629 96.8 14122976 java 24629-to-hex-6035
17668 27416 98.7 14122976 java 27416-to-hex-6b18
17668 28265 98.5 14122976 java 28265-to-hex-6e69
Can in be because the BTG filters are using the non thread safe HashMap ?