Great Article to review concepts -> Link
- You only need real or syntethic user monitoring
- You can model your environment manually.
- Monitoring software needs to reside in house.
- The Configuration Management database (CMDB) is a single physical repository of all knowledge.
- A Company’s infrastructure monitoring strategy can operate in its own detached silo. You need to be integrated to the business needs.
- Second or sub-second sampling rates are always necessary.
- Monitoring all the available metrics for a system or application is the best approach. Performance problems follow the Pareto’s rule: 80% of the problems are caused by 20% of the components of a system/application.
- All Systems in the IT enterprise will be monitored.
- Monitoring processes or services for an application suffices. It’s necessary to focus also on services interaction.
- Monitoring basic Infrastructure is enough. Architecture & Application are the main causes of performance problems.
Ok, this kind of reminds are very useful but I always have problems when I have to convince a customer spending some time reviewing performance problems that are not clear or they are in the background. What this kind of things happens I always try to:
- Clarify what is happening in an easy way (no technical words),
- List the benefits,
- List the technical actions that will be performed (a check list).
To finish, something that is very annoying is when different business use a shared service and anyone of these businesses want to get the responsibility of a shared performance problem. Where are the corporate responsibilities? Where is the organization?