perfkitbenchmarker/configs/dataflow_wordcount.yaml

#################################
# Worker machine configs
#################################
eight_core: &eight_core
  vm_spec:
    GCP:
      machine_type: n1-standard-8
  disk_spec:
    GCP:
      disk_size: 300

four_core: &four_core
  vm_spec:
    GCP:
      machine_type: n1-standard-4
  disk_spec:
    GCP:
      disk_size: 300

two_core: &two_core
  vm_spec:
    GCP:
      machine_type: n1-standard-2
  disk_spec:
    GCP:
      disk_size: 300

##################################################################
# Benchmark flags specifying Dataflow template and parameters
##################################################################
flags: &myflags
  dpb_service_zone: us-central1-a
  dpb_wordcount_out_base: <MY_BUCKET>
  dpb_dataflow_temp_location: gs://<MY_BUCKET>/temp
  dpb_dataflow_staging_location: gs://<MY_BUCKET>/temp
  dpb_job_jarfile: ./word-count-beam/target/word-count-beam-bundled-0.1.jar
  dpb_job_classname: org.apache.beam.examples.WordCount

#################################
# Benchmark variations to run
#################################
benchmarks:
- dpb_wordcount_benchmark: {
    dpb_service: { service_type: dataflow, worker_count: 1, worker_group: *eight_core },
    flags: *myflags
  }
- dpb_wordcount_benchmark: {
    dpb_service: { service_type: dataflow, worker_count: 1, worker_group: *four_core },
    flags: *myflags
  }
- dpb_wordcount_benchmark: {
    dpb_service: { service_type: dataflow, worker_count: 1, worker_group: *two_core },
    flags: *myflags
  }


#################################
# Alternative benchmark config examples
#################################
# dpb_wordcount_benchmark:
#   description: Run word count on dataflow
#   dpb_service:
#     service_type: dataflow
#     worker_count: 1
#     worker_group:
#       vm_spec:
#         GCP:
#           machine_type: n1-standard-4
#       disk_spec:
#         GCP:
#           disk_size: 300

# dpb_wordcount_benchmark:
#   description: Run word count on dataflow
#   dpb_service:
#     service_type: dataflow
#     worker_count: 1
#     worker_group:
#       vm_spec:
#         GCP:
#           machine_type: n1-standard-4
#       disk_spec:
#         GCP:
#           disk_size: 300
#   flag_matrix: cross_runners
#   flag_matrix_defs:
#     cross_runners:
#       dpb_dataflow_runner: [DataflowRunner, DirectRunner]