dont use a distributed tracing system right away.

hey everyone, ive been thinking about our monitoring stack. weve got metrics and logging covered, but im starting to feel like understanding request flows across multiple services is getting tougher. its not always obvious where a bottleneck is or what part of a distributed system is failing. has anyone found a good tipping point for when its time to actually invest in setting up something like Jaeger or Zipkin? what were the pain points that pushed you over the edge?