We're using SQL Server Enterprise 2008R2 on a large VM farm. We have hundreds of SQL Server VMs that all run very nicely.
However, when they were setup, the VM admins gave them huge amounts of memory - 64GB or 128GB. In most cases, they don't need nearly that much. The instances are all uncapped.
When I look at the instances, I see they're using 60GB+ of RAM...which makes sense, because SQL will use all the memory you give it for cache. If I look at its memory use, 80%+ is cache.
Is there a way to dig a little deeper and say "of that 80%, this percentage hasn't been touched in X hours/days" or "here is a breakdown of the age of cache by date and size"? I'd like to trim the amount of memory that these VMs and SQL instances use to recover RAM for the farm, but I don't want to just do it blindly.
I'm not the world's strongest SQL admin - these instances are administered by outsourced support who unfortunately is weaker than me :-(
