It would be valuable to be able to benchmark proposed remote write changes in a reproducible manner. Right now we are dependent on having a few people deploy branches to their instances to see if improvements work for them.

Perhaps have this optional so that remote write is not always exercised?