Currently the simulation holds a bunch of logs until a certain memory threshold is reached. Could there be a latency gain by sending logs every tilmestep since the FIRST tilmestep of data gets sent sooner? Or are there slowdowns since there is more slowdowns since updates to an individual file are not batched? Individual car based logging runs an expensive partition call then opens the corresponding file for the car and writes many timesteps in per car, but time series based logging just writes one for each tilmestep.
Should the sending strategy be coupled to the logging strategy since different logging strategies? Figure out how to implement this eloquently and maintain a strong separation of concerns.
Currently the simulation holds a bunch of logs until a certain memory threshold is reached. Could there be a latency gain by sending logs every tilmestep since the FIRST tilmestep of data gets sent sooner? Or are there slowdowns since there is more slowdowns since updates to an individual file are not batched? Individual car based logging runs an expensive partition call then opens the corresponding file for the car and writes many timesteps in per car, but time series based logging just writes one for each tilmestep.
Should the sending strategy be coupled to the logging strategy since different logging strategies? Figure out how to implement this eloquently and maintain a strong separation of concerns.