I'm looking for advice on how to use these distribution metrics/tags, to replicate something like a graph of the 95th percentile latency, for instance. The bucket_idx tags on count_per_bucket metric that result from opencensus' distribution aggregation seem rather difficult to utilize within datadog's graphs and alerts, as the bucket boundaries are defined in code and could also change over time.

Also, are there any plans for accepting opencensus' distribution data structure to have results more like what people are used to with the Histogram and Distribution types datadog intakes?