- Api calls to get rates takes around 2s - Connections from worker to mongodb is main bottleneck, takes around 1,5s - Optimize by adding KV as cache and read from KV instead of mongoDB