We have an http distributed repository server which we are also considering using as an agent handler. It would have strong connectivity to the database, so it's more of a question of whether this a good idea or not in general. Does anyone have experience for or against using this configuration?
There is no need to do this. The Agent Handler *is* already a repository. It is a proxy for your ePO server. When systems attempt to connect to the ePO server for updates, if they are connected to an agent handler, the agent handler intercepts the requests for ePO and then supplies the updates via the agent handler to the endpoint.
I would vote for not a great idea, point of repo's are to be close to the systems to provide updates. It doesn't need to be near the database at all.
Think of them as a cache for files, the agent handle is more for handling communication from agents to database.
I have a few desktops in each country setup as my distribution repo's (well the lazy cache option) to save WAN bandwidth.