Limits of "cloud infrastructure" monitoring?

[jeff_rowell]

I have been trying to understand whether there is any value in enabling full stack monitoring on Oracle and MS SQL DB servers (where no other application code runs on the servers). The documentation states that with "cloud infrastructure" mode there is “no application performance” monitoring. What I see, however, is that we see the processes that run on the servers and also see services. I can see that we would not have any "deep" monitoring of the services (given that the technology doesn't support that) but its not clear to me what else is "missing". Am I wrong is thinking that we also have monitoring of response time degradation, throughput and failure rate monitoring for these services?

[JamesKitson · Accepted solution]

The database services are actually detected on the application side (i.e. whichever process/service is calling the database) rather than on the database host. So as long as the calling application hosts are monitored in full stack mode you'll get the database service metrics like response time etc...

Infrastructure mode still gives you host health, process health, log info, and anything from plugins. The only thing you're really missing is services on that host which in the case of databases does not matter due to the reasons noted above. So generally you're better off leaving database hosts like this in cloud infrastructure mode as otherwise you're just wasting licenses.

[jeff_rowell]

Thanks James. I thought I had observer a situation in which a non-DB related service was showing up on a host that had only an infra agent, but perhaps I am wrong in that. At any rate, I think your answer clarifies things... although the DB service is being shown within the smartscape "on" the server where the DB runs it is actually being detected + monitored from the perspective of the calling application(s) that do have full stack agents.