> <i>Why Did So Many Startups Choose MongoDB?</i><p>Unanswered, & not even an attempt was made. There's a bunch of numbers at the beginning tracking it's meteoric rise & showing adoption, but nothing to discuss why.<p>I still wouldn't pick MongoDB but I still have huge huge huge respect for a rest-ful first datastore, for http+json as the lowest impedance way to read and write data. If anything it was just too easy to use, to get started with, and standards patterns/libraries/tools for data-control not adopted quickly nor widely enough.<p>The article mentions the rise of the MEAN stack (mongo express angular node) but doesn't really comment on what or why of this. But I think it iconifies how different things were with MongoDB versus before. What was so different was that, in some ways, one almost didn't need express/node: a well compiled website with apache httpd and mongo could in some cases be sufficient, or nearly sufficient to run a variety of apps. Because the database already spoke web, already stored & retrieved JSON documents. MongoDB was easy to learn, because it looked like the rest of the levels of the stack: it was about using the world-wide verbs we have to manipulate the data we have. Where-as most other databases, there is an impedance mismatch: we have to convert to other protocols, to other query languages, to other data-models.<p>I think there's still so much lost by having databases have their own protocols. There's neat attempts like PostgREST and pgrest & others to try to bring back some of this out-of-the-box web-native access, but overall it feels our expectations for how & where data gets stored still hinge around more special purpose protocols & means, are not tailored to the first two tiers of the three tier architecture that they still, to this day, anchor down.