Are confd subagents used in highly scalable systems?

I wanted to know whether, in a highly scalable virtualized environment will a single confd scale, if it has to provide configuration to 100s of nodes? Are subagents needed to support scalability? If there is a single configuration, can the master confd pass different subsets of the configuration to different nodes? How can confd be used in active/active architecture for scalability, not the mater/slave architecture used in HA.

Whether active/active is a better solution in large virtualized network, or it is better to use subagents?

With one active (master) - passing the configuration to multiple (subagents/other active confd) - is it possible to have standby (Confd) for these subagents. Basically, what we need is multiple pairs of active/standby nodes for a single large configuration


Have not received any reply yet. Can someone provide further details on this. If some examples are there, it would be really nice to follow them.