Load Balancing on sidecar

[ShapedTime]

I was wondering if load balancing can be done on sidecar level to send a cluster of gateway of collectors which will do tail sampling. I see that load balancer exporter uses consistent hashing and hashring mechanism for deciding on where to send the span so I am assuming many sidecar collectors, running on different nodes would still send the spans with same trace ids to the same backends but I couldn't find what is hashed for backends so like if I used a kubernetes resolver, would it still generate consistent results?