In a micro providers based mostly net back-end, a lot of the service’s (nodejs) comprises modules for deal with learn and write information seperately. When that exact service restarts it grabs the info from different micro providers and it remakes the learn mannequin cache (typically, a doc based mostly DB). And this service runs inside a docker-container.
On the subject of clustering or if merely talking when operating a number of occasion of the identical service, or when in a case which guide restart is required, this learn mannequin creation mechanism runs a number of occasions and it appears very dangerous follow. Particularly think about in a situation it must cache few hundred thousand entities each time, this impacts startup time, efficiency of that service.
What are the business greatest practices to create/handle learn mannequin information when service booting, restarting?