I’ve installed the 958 build and I can report that the out of memory errors don’t seem to occur as early as they did however It still seems to be throwing the error when there is still a reasonable amount of standby memory available. Ad there is still the issue of not being able to catch error when it’s thrown.
It seems to me that the power of a component like this is it’s ability to analyse a large set of data. My software for instance is looking at capacity and utilisation of a large number of windows servers, say around 5000 this could be several hundred thousand records. The 958 build has alleviated the issue a little but it still doesn’t go far enough. Is it possible to look again at the memory usage of AA, perhaps with a view to caching some data to disk to reduce the memory requirements. And as I mentioned in the pervious post, provide the ability to catch the error or raise an event when memory allocation changes so that we can monitor it and gracefully deal with the issue.