That works well in the days where you generally had on-site staff (or your datacenters were large enough to need onsite staff). We had remote POPs all over north America with perhaps a rack of servers in each location, and it made it much cheaper to send in an HP tech, essentially for free, than have someone on-staff flying everywhere when needed.
It all really depends on your use-case, I suppose. Nowadays most large datacenters are proficient enough that I think I’m fine with sending in spare parts and the service manual and letting their tech support have at-it. It also helps that unlike the HPUX days, nearly every server we have is redundant and able to be offline for a significant period if they mess something up 
