Henning Peter Jensen (10/26/2013)
Good idea - but you might want to adjust the max mem settings back to default if one of the two instances leaves the node. So if you add a schedule running hourly that resets the max memory to full capacity if the instances are running alone on the node, it will look almost perfect.
As soon as an instance is failed-back, so that each instance is each running on a separate node, these scripts/jobs will reset the max mem to their defined high values - in my examples to 245GB.