Amazing video. It’s bizarre to see a thoughtfully designed distributed system. The industry standard of solving scoped problems with 12 categories of overly generalised legacy data service clusters is maddening.
Surely you allocate memory each time a new account is created right? And hold all accounts in RAM? Apart from total number of transactions over all time (stored on disk I would think), it would be the number of accounts and current balance that would determine your RAM usage after startup right? That plus a pool of 10,000 structs or whatever in your batch size?
Same. Have slowly been going through them all these past few months. They do such a great job explaining complex topics, and finding simple efficient solutions. Brilliant. Considering watching their entire LSM stream next :)
@@yevgenydevine As an adult learning about interesting projects in your professional domain it’s healthy not to let your attention be entirely captured by the subjectivity of the 1min of humor