March 27, 2014 at 8:47 am
I was given the task to re-configure alert thresholds so that there are not too many unnecessary emails coming. In the company there were already configured in the third party tool, i see many emails coming in, but i am not exactly clear what is critical and should be taken action and whats not? how should i respond and who should i be contacting in case i need to take some action like killing a process that has been blocking other ones for an extended time.
Will appreciate your advice.
March 27, 2014 at 9:11 am
What you need to do is to figure out what is normal, and adjust the alert threshold accordingly. For example, it is normal to alert drive utilization at 80%, 90%, 95%. But let's say that you have a drive with one large database file that is sized for a couple of years data growth, and fills up 98% of the drive. So no need to monitor drive utilization, but you do need to monitor database space utilization. For drives with lots of database files, you need to keep free space, or a single batch load could blow the drive and cause many databases not to be able to grow.
If CPU spikes during nightly backup or batch processing, then 100% CPU utilization could be ignored for that time period if SLA's are being met.
So look at the 'top talkers', i.e. the items that generate the most alerts that are routinely ignored, and take care of them first. Repeat each month. From experience, I can tell you that when 99% of the alerts are false positives, the ones you need to pay attention to get lost in the shuffle. And that's when bad things can happen.
March 28, 2014 at 6:06 am
It really is down to a unique setting for each alert, sometimes a unique setting for each server for each alert. There's no single answer to any of these. As stated above, you need to look at the information over time to determine what "normal" is for your system.
Congrats on doing the right thing. Far too many people I've worked with tend to just turn off alerting rather than tuning the alerts.
"The credit belongs to the man who is actually in the arena, whose face is marred by dust and sweat and blood"
- Theodore Roosevelt
Author of:
SQL Server Execution Plans
SQL Server Query Performance Tuning
March 28, 2014 at 7:54 am
Thanks i appreciate your input, this gives me a good idea on how to analyze current alert trends.:-)
Viewing 4 posts - 1 through 3 (of 3 total)
You must be logged in to reply to this topic. Login to reply