I say near 100%, a lot of the dross comes from devices like UPS', which spam 3 emails a second.
Of all the alerts I receive, the quarter to half which we receive are from devices which are purchased and just attached to the network. This percentage is shrinking because of the actionable observability which we have enabled.
We don't use the alerting feature.
Before I started working with this customer the number was probably closer to the 75%. They were using very simple text only alerts and not purging very well. There were things monitored that didn't need to be and people just ignored the Ugly, and overly abundant alerts. I've created useful, attractive html alerts and minimized the duplications. The customer is much, much happier.
I have worked very hard to cut down on the irrelevant alerts, I did not like the 0-25%. I am under 5 %! Very proud of my deployment... it is small, so I have had time to work on the fine details.
We don't use alerting
We are at 99% or better, our goal is 100% and all of them are fed into our ticketing system. Sometimes something will slip through but people tend to complain when they are receiving incorrect tickets so they get cleaned up quickly.
Eh, it's getting there. The problem is the alert duplication that we're working through. 6 different alerts to say that something is down and using hard-coded emails, emails to individuals at that.
Thanks
We don't believe in dependencies at my workplace... also no one has to time to assign all our UPS. When we have power outages, it's a hot mess.
Inherited a basic "out of the box" install. Slowly working through it fixing stuff. Getting there.
I agree about the UPS spam. I switched to custom polling at short intervals on them to get less chatty alerts.
@zennifer is right getting to really good alerting is work but worth it.
Our company does not manage any on-prem servers or networking, so the mail SolarWinds tool we use is Service Desk. But it is interesting reading about the capabilities of other products.
When I'm building the alert environment, I start with all alerts OFF, and then after discussions only build alerts which the customer actually WANT, and that's after stressing that best practice defines all alerts as human-actionable, I suggest using reports.custom dashboards to capture anything less critical.It can be a really weird conversation to have, and frankly more than once I've had to accept defeat (the customer is always right, even when they aren't ) and build noisy alerts anyway...Also, good use of Custom Properties can leave you with only a handful of alerts, so use them well!
\m/
