Thank you Guenter.
As far as the end users that use APM, it has been a fight for the last five years to try to move the use from "Is it responding?" where the APM can be customized to be more of a infrastructure monitoring through the epagent to a performance management.
So my initial question was hopeful that I could try to get more eye candy with APM version 10 that I could attempt (again) the APM users view and use of APM.
Which people are accessing APM?
I've been looking around about the user profiles that were used to define the various new features within Team center to try to use them to help me categorize my end users. Past that:
APM Admins: building alerts, dashboards, ADS, renaming and moving things around a bit
WebSphere Admins - Is the servers taking traffic, is that traffic balanced when there is a problem
Development Team - as another source of information
What are they Interested in?
There is the rub, the majority of our current alerts are infrastructure. I actually go a "application average response time" alert within the alert structure but it's threasholds are so high and long, the only time it goes off is when the application is 100% unusable. The other alerts (cpu/memory/file space/agent connection status/netstat sockets in use) I should have been a dentist with the number of times I have tried to pull KPI from any of them.
We have two APM environment clusters, one for production and one for non-production. Also have a test cluster for testing patches/upgrades before we impact our non-production users.
Filter Out? Strange thing is, our end users want all the metrics possible so when something does go south they might have a chance to point at one or two different metrics and say that is the cause.
Do you have a naming scheme for servers, agents applications?
Servers - we just shifting to yet another naming scheme so we have feet into the "tv show of the time" nouns to some cryptic protein chain. The new naming scheme does have a p for production, t for test, d for development and s for stress, then there are like a handful more. But in our production cluster only production servers are reporting to our production collectors.
Agents - We have like four agent names for epagents and then the WebSphere and WebLogic agents, but we haven't broke that down into the multiple application hosting clusters.
Applications - seem to have no naming structure since you have to really know what it is before the application name in the triage map makes any sense or understand how each application relates to all the other applications. We currently have more than 250 applications.
The problem is there isn't enough constant known and stable naming within our server, agent, application but was hoping for something in APM 10 to help with that, knowing it will be another pushing a rock up hill effort.
Currently we have the java agents which gives us the front-end applications and backends (if they go through a known interface type (JDBC/EJB)).
We do not have CEM or SOA agent, or cross agent tracing enabled. With the upgrade to APM 10 I might be able to get the SOA agent and cross agent tracing tested out (if everything else works and I have the bulk of the month per environment to upgrade). From the sounds of it, that might make the team center better.
So, I'm going looking again for the end user profiles (Pete and gang) and see if I can align the various needs and skills to my end users.
Thanks again,
Billy