I am doing some R&D on ePO 5.0 and McAfee Agent 4.8. One of the new features in agent 4.8 is RelayServer. I am trying to figure out in what circumstances you would use this feature. There is little detailed information available, and the official release notes only cover very briefly what it does.
From the limited info that i have read i am presuming that when this feature is enabled, then local agent traffic will be diverted to this relay agent, which subsequently comminicates with ePO? The 2 things that confuse me most are;
1. If you had multiple relay agents enabled, then how do other agents know that they should use a specific relay agent?
2. I thought Agent Handlers were designed to facilitate this type of activity, so why would you use a relay agent and not an agent handler?
I've probably got the concept completely wrong, so would appreciate it if somebody could put me straight on where using this feature would be useful.
1. Not sure you get to choose tbh. Agent s using relay broadcast/multicast to find the nearest available relay afaik.
2. Agent handlers do that, but also do a heck of a lot more as they speak directly to Sql for example.
Agent handlers should be used sparingly as they are quite intensive. To an agent, an Agent Handler looks and behaves exactly like the ePO server itself.
So for usage, you might have sites or locations that are not directly connected to the ePO server but don't need an AH because there are not enough machines to justify it. There a relay may be perfect.
Basically relays only do the things other agents do - but remotely.
When you say content, do you mean things like DAT updates and so on? If so, then the relay agent does not cache these - it's not a repository, like a SuperAgent repository is. It merely passes the traffic on to the repository machine.
Correct, that's what I meant by content.
I can colocate a RelayServer and a distributed repository, correct? This would be the ideal "satellite" setup for a small site as far as I can see (i.e. a full Agent Handler isn't warranted due to number of agents at site, but to address routing challenges and be WAN-firiendly I need a RelayServer and something to make updates locally available).
I believe you should be able to do this - it's not something I've tried, though, but in theory you'd need a relay agent and a distributed repository.