The "three-legged stool" of performance is made up of memory, CPU, and disk I/O. Combining two of them can hide significant information.
DPA track most waits using the Microsoft wait statistics name except for Memory/CPU. What wait statistics are included in Memory/CPU or how is Memory/CPU waits calculated?
While there may be waits associated with obtaining and allocating memory, I need to know how much memory is being allocated.
One significant use case is for determining the cause of a sudden drop in PLE. If I look at the Timeslice tab and have PLE in the Resources tab at the bottom of the page, I can see when the event occurred that caused the PLE drop. I can then go to the SQL tab for that narrowed-down time period to see what SQL had significant CPU/Memory waits, but that doesn't help much. Nothing shows me a chart of the memory allocated for each of those SQL statements, which would likely make it obvious what the cause of the PLE drop was.
See related suggestions:
This was in Brent Ozar's Office Hours this week:
Brent Ozar: We got a follow-up on the memory question, and it is exactly as I suspected. There is a popular monitoring tool called SolarWinds, and this person says, “We’re using SolarWinds and memory CPU is listed as the metric.”
By all means what you want to do next is go talk to SolarWinds, talk to their support team, and ask what you should do about that wait. Maybe more specifically, ask exactly what that wait type means. It’s a very common question that we get amongst SolarWind users. It does not mean that you’re waiting on what you think it does.
