SQL Server Performance Forum – Threads Archive
High Memory usage.
Hi, I have a test Server having the following configuration : Model Type : DL360 G3Processor : 3.06GHz
Speed
Memory Size : 2GB
Hard Drive : 2 x 146GB (Raid 1) Obviously it is a low end server. To monitor the Server Performance BMC Patrol Monitoring Application has been installed. What has surprised me that the Application generates the following error message even during the least/no usage time : <Server> : Memory usage at 90.61% (/NT_HEALTH/NT_HEALTH/MemoryUsage). Although this is one time figure that I demonstrated it goes up even beyond 95%. Can you please advise the basic reason behind it. Arindam Ganguly
The Memory configuration in SQL is dynamic and it will only release when OS requests back, even we too get these alerts very often. So what we have checked is to ensure there are no performance issues or database issues at this time, then supress that alert on Patrol application. Also check whether any database maintenace tasks have been performed just before that alert is generated. Satya SKJ
Microsoft SQL Server MVP
Writer, Contributing Editor & Moderator
http://www.SQL-Server-Performance.Com
This posting is provided AS IS with no rights for the sake of knowledge sharing. The greatest discovery of my generation is that a human being can alter his life by altering his attitudes of mind.
let us know if you feel that Patrol actually provides meaningful purpose,
and weigh this against the alerts that do not result in any action
and any problems caused by Patrol same with other performance monitoring apps
Hi All, Thanks for your feedback. In fact I was also under impression that the Patrol provides some meaningful data at the very begining. But when observed it was found that alert is spasmodic and qurying further came to know that before each alert that Patrol submits, it experiences the exceeding of Memory usage threshold 5 times. It may even span for 4-5 hours. Thereby time and incidence ratio is at high end. I have also observed that it never provides the duration of the threshold overlimit. Even the Windows event log that I checked does not figure any abnormal incidence, neither I could get any Memory leakage issue. The performance wise too did not experience any bottleneck at the time of such incidence. I just got your view and thereby confirmed my conscent. Thanking you once again. Arindam Ganguly
it is more or less my opinion that almost all of the performance monitoring products provide nearly worthless value,
and are likely to be the cause of any problems it seems that some programmer were assigned the task to build a product,
who know how to look up API calls to collect information
but since they do not solve actual performance problems profesionally & scientifically,
all they do is ask other people what are acceptable values for certain counters. more on this later
]]>