I'm configuring SolarWinds for the first time on my environment. Currently I'm just completed adding my SQL database servers into monitoring & boom my dashboard showing reds on all my SQL databases, apparently due to physical memory utilization. High physical memory utilization is quite normal for any SQL database servers as SQL instance will manage it's own memory setting.
So is it a good choice to remove physical memory monitoring for SQL servers?
In my opinion you dont need to monitor the memory of a SQL Server. Usualy you configure your SQL Server to use 95% fix.
If you think to monitor the memory due to performance bottlenecks, I would rather recommend you to monitor the Page-Life-Expactancy (PLE).
This is the avarage time the SQL Server page can stay in memory, until it needs to be read again from disk.
If you have a bad PLE you usually should add memory to that SQL Server Instance.
In my opinion i tend to keep it, but reconfigure the warning and critical thresholds in line with what my DBA's advise. obviosuly they can allocate large chunks of memory for SQL jsut to consume, however they will also be aware that if other processes are running it could indicate a performance issue.
I think changing the warnign and critical threshold for memory should be adapted with relevant logic on there that says if it happens for X out of Y polls then it is a problem. That way you can alert off abnormlities you want to know about in a consist way to all other nodes.
It is too risky to have SQL hosts exempt from Memory monitoring, especially if they are business critical. If the SQL DBA's are allocating too much memory and it causes problems the monitoringwould indicate resource contention that needs managing.
SolarWinds solutions are rooted in our deep connection to our user base in the THWACK® online community. More than 150,000 members are here to solve problems, share technology and best practices, and directly contribute to our product development process.