I am not going to link to this article – no point in giving it more attention. The anonymous post claimed data loss and basic disaster using MongoDB. I call bullshit on this anonymous rant. Why was it posted anonymously?
I am sitting in an airport waiting to fly home right now: just finished extending a Java+MongoDB+GWT app and I am starting to do more work on a project using Clojure+Noir+MongoDB.
I do have a short checklist for using MongoDB:
- For each write operation I decide if I can use the default write and forget option or slightly slow down the write operation by checking CommandResult cr = db.getLastError(); – every write operation can be fine tuned based on the cost of losing data. I usually give up a little performance for data robustness unless data can be lost with minimal business cost.
- I usually use the journalling option.
- Use replica pairs or a slave.
- I favor using MongoDB for rapid prototyping and research.
- I use the right tool for each job. PostgreSQL, various RDF data stores, and sometimes Neo4J are also favorite data store tools.