Metrics

[jeffjunzhang]

Hi,
I'm currently testing Clipper with DockerContainerManager in a local cluster.

I'm using Prometheus server at localhost:9090 to check the metrics. There are multiple available metrics. Could we know the exact definitions of them? Is there any doc on this?

For example:
for throughput, there are: model throughput, app throughput and internal aggregate_model throughput, what are the differences?
Also, there are throughput_rate, throughput_rate_15min, throughput_rate_5min and so on.