MongoDB gets support for multi-document ACID transactions

But be certain not to conflate your respect as a business strategist with your judgment as a mindful developer. To speak clearly: By systematically playing a weak spot of ours [1] they have used countless of small teams as a stepping stone to sell their business contracts to large players, while hurting a lot of these small teams with an (at the time) inappropriate product for their needs. And through these huge costs they still made a product that is inferior to one that was designed properly.

As a community (both as a startup, as well as a developer community) we should resent these tactics and try to find ways to protect us against players that abuse the common good of mindshare. And lest you say, that is the price you have to pay to at all get a product like MongoDB in harsh business environments: We could also lobby for open source funds that are organized like research funds, producing fundamental technology that benefit everyone. Not every technology fits the model of for-profit startup innovation.

[1] Our community has very little defenses against marketing that comes from our midst, aiming to produce the (false) impression that a disproportionate amount of our fellows have evaluated the product and found it to be excellent. See https://www.nemil.com/mongo/3.html for a discussion about MongoDb specifically (HN thread: https://news.ycombinator.com/item?id=15124306 )

Nobody cares about people who are healthy all their life. But someone who suddenly realizes they need to eat better and exercise, and they do, they are applauded. They are defended too, if they go back to old ways. And so on...

- a JSON store that works well with NodeJS - a geospatial database (yea, it has what I need) that was "easier" to work with - a database that made it easier for me to change my schema (if you just throw data at it, garbage in, garbage out).

Over the years, I've followed development, and adopted new features to make my life easier.

- I was one of the people who were excited about full GeoJSON support in the 2.* days, because that's something I depended on. - I've tailed the oplog for as long as I remember (never needed Redis), and have been learning about change streams (announced in 3.6) with the hope of submitting a PR to Apache Beam to support them. - I adopted a lot of the aggregation framework (Asya Kamsky from MongoDB personally helped me a lot) - The last time I migrated data from MongoDB was when I turned on WiredTiger in 3.0.*

My little replica has been up for as long as I remember, the only time I have downtime is when restarting my server.

When I start a new project these days, I still go to Mongo, because what held true in its early days still holds. It's quick to get something started. Yeah, there's $lookups, transactions in future; I've just leveraged features as they become available. I have Postgres in my stack, which runs TimescaleDB and is a backing store for my Gitlab instance.

Would I use MongoDB professionally? It depends on the use-case. Over the past 2 years I've worked with Oracle, SAP HANA, Teradata, Hive+Impala, etc. From OLTP to OLAP, but once in a while when I find it quick to, I still use MongoDB, and when I have it my way, I don't later migrate elsewhere.