in one of our datacenter facilities with multiple cages i.e. hundreds of servers and networking gear to support, the onsite work is less than 10 hours a week.

in one of our datacenter facilities with multiple cages i.e. hundreds of servers and networking gear to support, the onsite work is less than 10 hours a week. almost all failures are engineered against. one of our co-founders just goes there one afternoon per week since he lives nearby. he says it's nice to get out of the office occasionally. it's basically just rack/stack/wire/replace. as we move into more software automation the job will be less hours as failures will no longer be time sensitive. maybe one visit every month. i've never personally seen a robot in a datacenter so i'm not sure how the physical aspects are being automated but it's fairly low wage work so i figure it will take a long time before putting robots in DCs is cost effective. having said all that, it still requires someone smart onsite to know wtf is going on when something is seriously broken or needs to be built right the first time around.