We're using datomic in production. It's had its ups and downs. For one, having raw data available at in-memory speeds really changes the level of expressiveness you have in your code; you no longer are constrained to packing every question about your data into a giant query and sending it off - you can instead pull data naturally and as needed. Many of our queries make multiple queries and are high performance.

The licensing is a huge pain in the ass. If I accidentally launch an extra peer over our license limit, our production environment will stop working until the extra peer comes down. This is really butting heads with the growing popularity of abstracting physical servers as clusters so I think the strategy is kind of a mistake on cognitect's behalf.