[Feature Request]: Support more latency, throughput, and resource usage metrics in the benchmark framework

[frank-shuo-chen]

What would you like to happen?

The Java SDK benchmark / load test framework under it/google-cloud-platform/src/main/java/org/apache/beam/it/gcp/ supports a limited set of metrics by default. Support for the following metrics are needed to enable richer, more informative benchmarks:

• Per stage E2E latencies at various percentiles
• Input throughput normalized by vCPU time
• Resource usage and cost estimation (vCPU, memory, etc.) over a set, steady state interval rather than over the entire lifetime of a streaming pipeline

Issue Priority

Priority: 2 (default / most feature requests should be filed as P2)

Issue Components

• Component: Python SDK
• Component: Java SDK
• Component: Go SDK
• Component: Typescript SDK
• Component: IO connector
• Component: Beam YAML
• Component: Beam examples
• Component: Beam playground
• Component: Beam katas
• Component: Website
• Component: Infrastructure
• Component: Spark Runner
• Component: Flink Runner
• Component: Samza Runner
• Component: Twister2 Runner
• Component: Hazelcast Jet Runner
• Component: Google Cloud Dataflow Runner