Back in 2012, my former boss forced me to use marklogic as a database to ingest and help display positional data in real-time. Unable to get marklogic to work without building fairly complex caches around it, I then decided to look around for a database, which would have performance properties of a cache. With none found (I missed redis somehow) and no money to spend on KDB, I started to cobble together file based storage and cache to work as a cohesive unit. Let me tell you, I learnt a lot more in that one year than my prior 15 years developing software.<p>Addicted to learning new things, I doubled-down on turning the storage+cache idea into a functioning database, without realising how hard it was to turn this idea into reality. Unwilling to give up, and fast forwarding to almost 2020, I'm asking you to provide feedback about what I've accomplished so far. QuestDB is currently usable as a single node or embedded, whilst not yet supporting horizontal scalability (on our roadmap in 2020'). Thank you for your time, I look forward to hearing your honest feedback.