37signals is completing its on-prem move

(theregister.com)

57 points | by xrayarx 2 days ago ago

8 comments

  • jdboyd 2 days ago ago

    They say that the move didn't require adding any staff, but aren't all of these physical servers going to require somebody to do occasional physical tending to them? What happens when a server fails? Sure the software might be able to handle failover and absorb the load, but isn't somebody going to have to go deal with the failed server at some point?

    • adamcharnock a day ago ago

      This can be mostly handled in the planning. Plan for server failures, but datacenters will also offer 'remote hands'. You can store a bunch of spare parts in one of your racks (see rack-mount-drawers), then call up the DC, and ask them to swap component A in server B with spare C in drawer D.

    • nchmy 2 days ago ago

      Dhh has addressed this a dozen times - they shipped the servers to some datacenter which manages everything for them. And they, obviously, have hardware redundancies

      • tehlike 2 days ago ago

        Colocation, essentially.

    • izacus a day ago ago

      Colocation services are decades old and exist all over the world. Did you never hear of them?

    • t312227 a day ago ago

      hello,

      imho. as always ...

      its not only colocation of (server|network|...)hardware.

      its possible to outsource your infrastructure to a business which "manages" it for you.

      idk. think ibm, but also many other larger & smaller vendors who offer such services :)

      just my 0.02€

    • porridgeraisin a day ago ago

      Even if they don't use colocation, is there a business difference between hiring dozens of AWS engineers vs contracting dozens of infra engineers?

  • ChrisArchitect 2 days ago ago