When rolling new changes to the osquery configuration of nodes, it would be a good idea to implement a system to shard the configuration so it will not get to all the nodes at once.