While json fields in traditional SQL are useful in many ways (avoiding that infamous "extra" column) it really doesn't compare to mongodb. The aggregation fram

While json fields in traditional SQL are useful in many ways (avoiding that infamous "extra" column) it really doesn't compare to mongodb. The aggregation framework in mongo for working with these documents is very powerful and if you run pg json fields to do anything other than retrieve extra fields (ie: if you try to query/aggregate it's fields meaningfully as you would a standard field) you'll find they fall over pretty fast with modest amounts of records. The simple scalability of mongo (sharding/replication) is something that, while the recipient of much mocking initially, now at 3.6, gives enterprise level availability and scale with minimal oversight or expertise. I'm really liking the features in 3.6 (over the wire compression, replication hardening, streaming query subscriptions to name a few) and every release packs a ton of features and improvements without breaking anything. Postgre is a great relational database (one of my favorites) but its json field (which even mssql sports nowadays) is not a stand-in replacement for mongo in anyway way.