Grid control design and agent deployment

Question

To give you a brief overview of the setup here, it is currently an agent per instance instead of an agent per host. This was done to allow an agent to be moved along with an instance to another host in a cluster (we are using IBM AIX LPAR, and Sun Clustering) thereby allowing a seamless transition between hosts and history retention for the instance. We would like to change this to an agent per host, however, we need to keep the historical information of the instance should it change hosts. So my questions are:
1. Is it possible using some sort of script or process to have an instance/target move between hosts and retain its previous historical information?
2. Is it possible to have a single, summarized availability report for all instances? The report per instance works, however, doing a single summary report for all instances does not display the correct information (it only shows PIT, i.e. the current state of the instances not historical over x period of time).
Any help is appreciated.
Thanks.

Answer 1

Let me provide some details as to the environment as to the deployment which I am doing in a HA environment:
- There is a physical 'myhost1' with virtual hosts names oradev1, oratst1. Note I am not referring to virtual machines, just virtual host names, or virtual IP.
- Each VIP hosts a listener and database instance. For example, oradev1 -> lsnr-oradev1 + orad1, and oratst1 -> lsnr-oratst1 + orat1
In this design each database and listener can be moved to another physical host separate from the other, say the entire oradev1 stack moves to 'myhost2'.
How would the agents be deployed in such a design? I think an agent per physical host, with the VIP used to monitor the virtual stacks, such has as each virtual host containing the listener and database instance. This way when a virtual stack moves we can just switch the agent monitoring that stack to the new host and retain the history of the targets.
Is my thinking correct? Is there a better way to retain the history when moving the virtual stacks perhaps doing an agent per instance instead of physical target?
I hope this provides more details. Thanks.