hi
simply asked: should i have an all in one jsonl in /dev/shm no matter what?
more complicated, i understand that for example having 1 jsonl for each symbol is a definite IO no no.
but i feel like watching 250 symbols in peak times (9:30AM ET, 4:00PM ET), would make the all-in-one jsonl VERY crowded. somehow i feel like having separate jsonl files for every 40 or 50 symbols can be handled better without missing prints.
i only have experience for now watching T:t AND T:q so i may just want to ditch T:q for now and see if trade prints only are less crazy. (for now i don’t have a use for T:q. may will need it later for other modules/calculations/whatever though.)
still. peak hours, monitoring the possible runners i could easily have 10+ million prints a minute. now that is insane. i have a better chance normalizing / load balancing them by splitting the watched symbols to 40-50 a piece buckets. i feel.
but please do share your thoughts.
thanks
Gabe