I've been nothing but an 'ops guy' for 20 years now, and I've been enthusiastically using MongoDB for my 'devops' work since 2009.

Having said that, I've never used MongoDB 'on the large'. I've always used MongoDB against datasets that mostly fit into memory, and use some other solution for 'bigger data'.

I suspect most of the MongoDB dislike is rooted in three points: 1. The 'Mongo People' really over-hyped it early on. 2. The decision to not have single-server disk safety (but providing very elegant multi-server data safety) was a poorly understood decision. I personally agree with their direction. Your data isn't really, really safe until it's on more than one system. 3. Based on what I've read, I don't think MongoDB scales up (even with their built-in sharding) as well as some other solutions. Perhaps this is limited to some kinds of use cases, and not others.

In my opinion, for people who do day to day fast development with dynamic languages, like myself, MongoDB, used correctly, is really fantastic, at least for many use cases.

really? all the ops people that I've worked with have been pretty enthusiastic about it, more so than a mysql instance with bugs in 5.5 that created huge performance issues with multi-table joins (i.e. http://dba.stackexchange.com/questions/13817/mysql-5-5-stran...)