DistributionSamples is similar to Distribution, but it lets the client deals with the sampling, rate is passed to the agent and not used for further sampling. WithMaxBufferedMetricsPerContext is ignored when using this method.
Because this is a shift compared to how other methods are behaving, this method is provided only in ClientDirect which provides direct access to some low level dogstatsd protocol features like https://github.com/DataDog/datadog-go#extended-aggregation
This is recommended in high performance cases were the overhead of the statsd library might be significant and the sampling is already done by the client.
DistributionSamplesis similar toDistribution, but it lets the client deals with the sampling,rateis passed to the agent and not used for further sampling.WithMaxBufferedMetricsPerContextis ignored when using this method.Because this is a shift compared to how other methods are behaving, this method is provided only in
ClientDirectwhich provides direct access to some low level dogstatsd protocol features like https://github.com/DataDog/datadog-go#extended-aggregationThis is recommended in high performance cases were the overhead of the statsd library might be significant and the sampling is already done by the client.