3 ms·

Can you speak to the architecture at all? e.g. an event hits Freshpaint, and 3 months later is is qualified as a defined "event" to be forwarded to 3 destinatio

Can you speak to the architecture at all? e.g. an event hits Freshpaint, and 3 months later is is qualified as a defined "event" to be forwarded to 3 destinations - how does it happen? I ask because I saw your previous comments and saw that 1) you're an expert at optimizing PostgreSQL, and 2) you said that Heap was harder to build because of the queries that users can define in the frontend. So basically, are you able to leverage PostgreSQL to build something like Freshpaint? I would imagine it is prohibitively expensive to store your event data in PostgreSQL if it's sole purposed is to be categorized and forwarded to other destinations, vs. being queried like it was in Heap. Amazing work and best of luck!

polote 7y ago
They probably store the XPath of each click, and when an event with the same XPath is created they browse through all the past click with this specific XPath
- slap_shot 7y ago
  Of course. But my question pertains _how_ the event data is stored. Heap stored the event data in Citus, and OP worked to optimize their PostgreSQL for queries across the data. Freshpaint doesn't query the data (in the aggregate form that Heap does), it simply needs to find events that qualify as defined events and forward them to their destinations. I'm just asking if PostgreSQL is still the right tool for that.