At Pushnami we were looking at several alternative databases that would support following architectural requirements:

• very quick prototyping for an unknown domain
• ability to support large amounts of data
• native ability to replicate and fail over
• full stack approach for Node.js development After careful consideration MongoDB came on top, and 3 years later we are still very happy with that decision. Currently we keep almost 2TB of data in our cluster, and start thinking about sharding.