[ https://issues.jboss.org/browse/TEIID-3583?page=com.atlassian.jira.plugin... ] Steven Hawkins commented on TEIID-3583: --------------------------------------- This issue should probably focus on the end requirement, which is heap throttling. There are a couple of considerations/clarifications needed. First is that we make no hard guarantees on heap usage - parameters like these can only influence the pattern of consumption and are based upon estimates (and of course the vm will still accumulate garbage well over any target). Second we may need to clarify what they mean by active throttling - are they looking to adjust these parameters at runtime (not currently possible) or even down to a query level (not currently possible). Third as mentioned above the affect will largely be dependent upon workload and how the other values have been set. Under a high uniform query volume reducing the max-reserve-kb may not influence memory consumption very much if max-processing-kb / max active plans allow for a large memory usage. Is it possible that we can get a better understanding of their goals? Also some of the work that was done for TEIID-2429 mitigates the need to adjust these values. If max-processing-kb is left at its default, then we'll automatically assume a balanced allocation plan such that we'll nominally not exceed max-reserve-kb by just plan processing. -- This message was sent by Atlassian JIRA (v6.3.15#6346)