Monitoring Oracle

Although Enterprise Manager Cloud Control needs a separate license to use it, it is probably one of the best utilities to monitor your estate. Strangely, at one point it needed an 11g Oracle database as a repository and was not compatible with 12c. You can use a 12c database now but not on Oracle Linux version 7.

https://oracle-base.com/articles/12c/cloud-control-12cr5-installation-on-oracle-linux-5-and-6

I use Oracle-Base as a chef would use a recipe book. Follow the instructions and hope you produce something tasty. So rather than put out the same recipe I will point you to the best cook books.

The great thing about Cloud Control is that you can put an agent on a server and monitor what is going on, RAC, Oracle Application Server Farms etc. It is not only Oracle you can monitor, Microsoft SQL Sever too, so a very useful application if you can have access. An alternative would be something like Quest Spotlight but I have only ever been on one site that had this running and that was a few years back so if someone has any recent experience with this please let me know.

https://www.quest.com/video/overview-of-rac-in-spotlight-on-oracle891910/

If you don't have Cloud Control or Spotlight there is still the 12c Database control aka EM Database Express but it is built inside the database server so limited functionality. SQL Developer has some tools for monitoring but most of the time you are back to manually checking the alert log and trace files for problems, running O/S tools like prstat, vmstat, top etc. 

Sys Admin guys can be useful for problems at the O/S level (especially for h/w issues - "dodgy" disks can be a pain) and speak to network teams (firewalls, routers, switches can be a pain). If you have the diagnostics pack run AWR reports when you are seeing issues (remember AWR reports are different for stand alone instances and for RAC).  Most DBAs have their own scripts and if you have been running a system for any length of time you get a "feeling" for what is going on. 

Oracle monitoring is not just a one approach fits all it you often need a holistic approach. I remember after moving a Sun Server the performance was terrible. The only difference was the network had changed so the network lads were getting the blame. After a week I spotted the memory was only half of what it should have been. After a power down and a re-seat of the memory the server was back to what it was. I have learnt that the "Occams Razor" quite often applies when it comes to Oracle.