Legacy HDS Forums

Tuning Manager Agent Architecture

Discussion created by Legacy HDS Forums on Aug 26, 2012
Latest reply on Sep 3, 2012 by Legacy HDS Forums

Originally posted by: Optic


I'm trying to sort out exactly how Tuning Manager works from an architecture point of view.  Yes - I've read a lot of the manual but I find that it tends to get very technical without giving an overview of how it works.

If my questions are covered in the manual please point me to the right page.  Maybe I'm just being stupid and missing it.

I'm interested in how Agents work with Tuning Manager.  More specifically - lets use an "Agent for RAID" example talking to a USP.

Agent for RAID is installed on a machine which has a Command Volume presented to it.  An "instance" is created (type 2) and pointed to the CMVol.  The "test" tools all report that everything is OK with the instance.  Agent is pointing at the HTnM server.

How does the flow of performance data work?  Specifically, these are my questions:

1. How often does Agent for RAID connect to the Command Volume and "pull" performance data?
2. When it does this, where does it put the data? (Eg, does it go straight to HTnM or does it go to the local data store that the Agent seems to create).
3. If it goes straight to HTnM, what is the purpose of the "polling schedule" in HTnM?
4. If it goes to the local data store on the Agent machine, what mechanism causes this data to eventually make it's way to HTnM?  Maybe this is what the polling schedule is?
5. If I'm on the wrong track, what does the "polling schedule" actually do?  Does it reach out and contact the agents etc?

Thanks in advance!