Today’s information processing and database systems are more capable than ever. Nowhere is this so apparent as in the world of NoSQL databases, now mainstream in enterprise software architecture. With such a dynamic store of data, business exploration and organic growth happens naturally and in production, thanks to the absence of an always enforced, rigid schema.
But hang on… do you know the data in your system?
Large software systems are constantly evolving (and we all love hot code updates). However, data once produced might stay there forever. Not surprisingly, accessing data written by an older software version may lead to unpleasant discoveries. You might want to check your assumptions before pushing new code to production… or you might want to validate the data, cleaning up artifacts of an old bug that has been fixed long ago. Maybe you are just curious and would like to learn more about your data.
Dumpsterl is my humble answer to these challenges. Dumpsterl scrutinizes the contents of an entire table, a single column, or an arbitrary stream of Erlang terms. It goes through the values (or a random subset of them) and builds comprehensive metadata, essentially a specification of the data encountered. While doing this, it collects representative samples possibly annotated by key and timestamp to support further probing. Dumpsterl is flexible and easily extensible, so you can feed it with virtually any source of data.
You might be interested in my talk Dumpster Dive your Erlang data! that I gave at the Erlang User Conference in 2017 to introduce and showcase dumpsterl. The handouts of the talk might also be useful if you don’t want to watch the video – along the slides, they also contain an edited transcript of what I was telling the audience. (Although you will miss the most interesting part, the live demo, which was not slide-based.)