Use Case

As a... (who is this for)
I want... (what is the desired change)
So that... (why is the change desired)

Dev Experience

• Allows for easier grouping of benchmarks into performance contexts

Dependencies

• None

Risks/Unknowns

• Will the range of values for the normalized throughput be easy to work with or is there a better way to compute it?

Acceptance Criteria

Implementation Requirements

• Update primes benchmark task size to bring primes benchmark result to 10^0 order of magnitude to make normalized throughputs easy to work with.
• Mark the following benchmark results with the alerting-benchmark tag
  • ping
  • aggregateAMillionDocumentsAndToArray
  • aggregateAMillionTweetsAndToArray
  • findOne
  • findManyAndToArray
  • findManyAndEmptyCursor
  • smallDocInsertOne
  • largeDocInsertOne
  • smallDocBulkInsert
  • largeDocBulkInsert
  • ldjsonMultiFileUpload
  • gridfsMultiFileDownload
  • gridfsMultiFileUpload
  • gridFsUpload
  • gridFsDownload
  • runCommand
• Mark the following benchmarks with the spec-benchmark tag
  • findOne
  • findManyAndToArray
  • findManyAndEmptyCursor
  • smallDocInsertOne
  • largeDocInsertOne
  • smallDocBulkInsert
  • largeDocBulkInsert
  • ldjsonMultiFileUpload
  • gridfsMultiFileDownload
  • gridfsMultiFileUpload
  • gridFsUpload
  • gridFsDownload
  • runCommand
  • singleBench
  • readBench
  • writeBench
  • parallelBench
  • multiBench
  • driverBench
  • bsonBench
• Add a new measurement to the ping benchmark called normalized_throughput that is calculated as (throughput of ping benchmark in megabytes per second)/(throughput of primes benchmark in megabytes per second)
• Add a new measurement to all other benchmarks called normalized throughput that is calculated as (throughput of benchmark in megabytes per second) /(throughput of ping benchmark in megabytes_per_second)
• Add relevant tags to each benchmark
  • cursor-benchmark
  • read-benchmark
  • write-benchmark
  • alerting-benchmark

Testing Requirements

• Manually verify in CI that the tags have been added correctly
• Manually verify in CI that the normalized_throughput values are being added correctly

Documentation Requirements

• Add test readme note about the perf contexts and tags we use

Follow Up Requirements

• None