Moved to epo community
Re: "What is the difference b/w agent handler and ePO server."
From the McAfee Knowledge base, article 'PD22508 - ePolicy Orchestrator 4.5 Agent Handler White Paper' this is still relevant to that question:
"The ePolicy Orchestrator (ePO) 4.5 server separated out the services of the server dealing with agent requests, and added management logic to allow deployment of instances of these services.
This new component of the ePO infrastructure has been termed an Agent Handler.
In ePO 4.0 and previous, there was a single ePO server that agents could connect to and receive policy and task updates. Since the ePO server was responsible for handling every agent connecting to it, there was a limitation on the deployment size an ePO server could handle.
The only option to increase the scalability of a single ePO server was by moving the database out. Otherwise the ePO server could be scaled vertically (through bigger and faster hardware) instead of horizontally (through more servers to distribute the load). The introduction of Agent Handlers in 4.5 gives customers the ability to grow their logical ePO infrastructure horizontally adding multiple Agent Handlers to scale agent connectivity."
Basically it's the same as an ePO server but without the event parsing.
The ePO server is also an Agent Handler by definition.
Agent Handlers have pretty much the same system and network requirements as an ePO server - especially in terms of the need for excellent connectivity to the Sql server.
The same white paper discusses usage scenarios.