I came across an unusual scenario on an instance where only half the physical RAM was being utilized but I was observing PLE values between 1-3 seconds constantly.
Windows Server 2003, Enterprise 5.2.3790 (32-bit) SP2
4 GB RAM
2 Xeon 2.3 GHz CPU
SQL Server 2008 SP2 (10.0.4000)
Min/Max Server Memory were set at defaults and the most resource intensive query was the backup history check by Quest Spotlight. Only 1.6 GB of RAM was utilized, 1.5 of which by buffer and procedure cache.
After setting Max Server Memory to 3 GB, PLE started to climb and is currently at 21 minutes.
I am wondering why Max Server memory would have this positive impact on PLE when less than half the onboard RAM is currently utilized. My first thought is O/S memory trimming was in play before the change in the setting but how can I confirm this?