Isn't it more significant at smaller scale? That is, if you have less computers running to serve requests, the downtime of the singular system will be more pro

Isn't it more significant at smaller scale? That is, if you have less computers running to serve requests, the downtime of the singular system will be more pronounced (as opposed to rebooting one machine out of 20 in a rack).