Parse

» see parse.com as it lived, on the wayback machine

~ the goodbye ~

parse was a backend you did not have to build, and for a few years that was the entire pitch. i was a computer engineer watching mobile development from one desk over, so i understood the appeal immediately. you wanted an app that saved user data to the cloud and pushed notifications, and normally that meant standing up your own servers and writing all the boring plumbing yourself. parse said: don't. drop in our sdk and ship the thing. for a working developer that was close to magic.

parse came out of y combinator in 2011, founded by ilya sukhar and a small team. for whatever reason it grew fast. by 2012 it had something like twenty thousand developers on it, growing forty percent a month, and a couple of years later it sat under hundreds of thousands of mobile apps. facebook bought it in april 2013 for a reported eighty-five million dollars, which at the time looked like facebook getting serious about being the backend for everyone else's apps.

that is not how it went. facebook never really turned parse into the platform play people expected. on january 28 2016 the company announced it was winding the hosted service down, with a hard cutoff one year out, by then around 600,000 apps were running on it. to be fair, it did not just delete everything. it open-sourced parse server the same day, so teams could in theory move their backend onto their own infrastructure and keep going. but a year is not long to re-home an app you built on the assumption someone else would run the servers forever.

~ the chronicle ~

~ the breakout ~

the thing parse got right was timing. back then. in the early 2010s every company suddenly needed a mobile app, and most of them had no idea how to run server infrastructure. parse sold the part nobody wanted to build. you got cloud data storage and push notifications behind one sdk, billed by usage, and for a small team that was the difference between shipping in a week and shipping in a quarter.

facebook's purchase made it look permanent. who walks away from infrastructure that hundreds of thousands of apps depend on? for a while parse was the default answer when a developer asked where to put the backend, and that default is exactly what made the shutdown sting.

~ the actual end ~

the shutdown was handled better than most. i was not following the exact mechanics at the time, but the shape of it was clear enough. facebook gave a full year of notice and released parse server as open source, so the api you had written against did not have to vanish. teams with the engineering muscle stood up their own parse server on aws or heroku and carried on. teams that did not, the small shops and solo developers who picked parse precisely so they would never have to touch a server, were left scrambling to rewrite their entire backend before january 2017.

~ the next chapter ~

parse did not fully die, which is the strange part. parse server is still out there as an open-source project, maintained by a community well after facebook stopped caring, so the code outlived the company. but the lesson the rest of us took away was harder than the code. a backend-as-a-service is a landlord, and the landlord can leave. for whatever reason a generation of mobile developers got quietly more skeptical about building their foundation on someone else's hosted promise, and you can draw a line from that skepticism straight through every argument people later had about firebase and the managed backends that came after.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »